Apuntes_ML.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Apuntes_ML.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "mount_file_id": "1lSTJGulWeLAR3_h0Ck0h_2lHlhkSKMYx",
      "authorship_tag": "ABX9TyP4M2z51y/RIXTXPKjskCIF",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mavivi95/Notebooks/blob/main/Apuntes_ML.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RVfC0fWLj96t"
      },
      "source": [
        "##Practicas codigos Python y Machine Learning"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Aan-4cLqkhzC"
      },
      "source": [
        "Para comenzar cargar las librerias de Python necesarias en las técnicas de Machine Learning. Las librerias y clases importadas en el orden del código del bloque siguiente son:\n",
        " \n",
        "\n",
        "* **csv**: para importar datos en formato csv\n",
        "*   **Numpy**: para el trabajo con datos matriciales\n",
        "*   **sklearn.model_selection**, función **train_test_split** para segmentar de manera adecuada el conjunto de datos en entrenamineto y prueba\n",
        "*  **sklearn.preprocessing**, función **SrandadScaler** para estandarizar los datos y evitar problemas con unidades desbalanceadas\n",
        "* **sklearn.linear_model**, función **Perceptron** para establer la clase del perceptron. Esta clase tienes los metodos para entrenarlo y analizar su rendimiento\n",
        "* **matplotlib.plot** libreria empleada para realizar gráficas en python\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C5OyNRHbkgq0"
      },
      "source": [
        "import csv\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.linear_model import Perceptron\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DT2auxfy66AV"
      },
      "source": [
        "from matplotlib.colors import ListedColormap\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "\n",
        "def plot_decision_regions(X, y, classifier, test_idx=None, resolution=0.02):\n",
        "\n",
        "    # setup marker generator and color map\n",
        "    markers = ('s', 'x', 'o', '^', 'v')\n",
        "    colors = ('red', 'cyan', 'lightgreen', 'gray', 'blue')\n",
        "    cmap = ListedColormap(colors[:len(np.unique(y))])\n",
        "\n",
        "    # plot the decision surface\n",
        "    x1_min, x1_max = X[:, 0].min() - 1, X[:, 0].max() + 1\n",
        "    x2_min, x2_max = X[:, 1].min() - 1, X[:, 1].max() + 1\n",
        "    xx1, xx2 = np.meshgrid(np.arange(x1_min, x1_max, resolution),\n",
        "                           np.arange(x2_min, x2_max, resolution))\n",
        "    Z = classifier.predict(np.array([xx1.ravel(), xx2.ravel()]).T)\n",
        "    Z = Z.reshape(xx1.shape)\n",
        "    plt.contourf(xx1, xx2, Z, alpha=0.3, cmap=cmap)\n",
        "    plt.xlim(xx1.min(), xx1.max())\n",
        "    plt.ylim(xx2.min(), xx2.max())\n",
        "\n",
        "    for idx, cl in enumerate(np.unique(y)):\n",
        "        plt.scatter(x=X[y == cl, 0], \n",
        "                    y=X[y == cl, 1],\n",
        "                    alpha=0.8, \n",
        "                    c=colors[idx],\n",
        "                    marker=markers[idx], \n",
        "                    label=cl, \n",
        "                    edgecolor='black')\n",
        "\n",
        "    # highlight test samples\n",
        "    if test_idx:\n",
        "        # plot all samples\n",
        "        X_test, y_test = X[test_idx, :], y[test_idx]\n",
        "\n",
        "        plt.scatter(X_test[:, 0],\n",
        "                    X_test[:, 1],\n",
        "                    c='',\n",
        "                    edgecolor='black',\n",
        "                    alpha=1.0,\n",
        "                    linewidth=1,\n",
        "                    marker='o',\n",
        "                    s=100, \n",
        "                    label='test set')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NVJn28Ony5Ee"
      },
      "source": [
        "###Calidad del Vino Rojo\n",
        "\n",
        "Los datos corresponden a 1599 muestras del producto \"Vino Verde Tinto\" que incluyen 11 características fisicoquímicas del producto y una columna de etiquetas donde se calificó por un experto la calidad del vino en una escala de (0-10) puntos, siendo 10 la mejor calidad.\n",
        "\n",
        "* Acidez fija\n",
        "* Acidez volátil \n",
        "* Acido cítrico\n",
        "* Azúcar residual\n",
        "* Cloruros\n",
        "* Dióxido de azufre libre \n",
        "* Dióxido de azufre total\n",
        "* Densidad\n",
        "* pH\n",
        "* Alcohol"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Hbc6FtkEj9Cx",
        "outputId": "b311fe68-1153-4826-af1a-989e6064ad22"
      },
      "source": [
        "from numpy import genfromtxt\n",
        "data_with_header = genfromtxt('/content/drive/MyDrive/Dataset/winequality-red.csv', delimiter=';')\n",
        "data = data_with_header[1::,:]\n",
        "\n",
        "x = data[:,0:-1]\n",
        "y = []\n",
        "for label in data[:,-1]:  y.append(int(label)) \n",
        "#data\n",
        "print('Numero de clases - Calidad vino rojo:', np.unique(y))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Numero de clases - Calidad vino rojo: [3 4 5 6 7 8]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "q_RY8mijsurU",
        "outputId": "e13ec6c2-42b0-477d-8394-8aa230fa0273"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.3, random_state=1, stratify=y)\n",
        "print('Datos en y_train:', np.bincount(y))\n",
        "print('Datos en y_train:', np.bincount(y_train))\n",
        "print('Datos en y_test:', np.bincount(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos en y_train: [  0   0   0  10  53 681 638 199  18]\n",
            "Datos en y_train: [  0   0   0   7  37 477 446 139  13]\n",
            "Datos en y_test: [  0   0   0   3  16 204 192  60   5]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YQR72sSHzIr9"
      },
      "source": [
        "Como se puede apreciar, las clases estan desbalanceadas inicialmente no hay calificaciones de 0 a 2 en calificaciones de la calidad, hay muy pocos datos calificados en 3 y muy pocos para 8. Por esa razon se creará un nuevo vector de etiquetas, los valores menores o iguales a 6 crearan una clase y los valores mayores la otra."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0rtVTRhlyxgO",
        "outputId": "bbe95004-da18-4aeb-d820-4666c8c4ff85"
      },
      "source": [
        "y_binary = []\n",
        "for label in data[:,-1]:  \n",
        "  if label >= 6: y_binary.append(0)\n",
        "  else: y_binary.append(1)\n",
        "print('Numero de clases - Calidad vino rojo:', np.unique(y_binary))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Numero de clases - Calidad vino rojo: [0 1]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TG47z-fB0yNM",
        "outputId": "d7db0a66-7591-4243-87df-869c50e5d731"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(x, y_binary, test_size=0.3, random_state=1, stratify=y_binary)\n",
        "print('Datos totales por etiqueta:', np.bincount(y_binary))\n",
        "print('Datos en y_train:', np.bincount(y_train))\n",
        "print('Datos en y_test:', np.bincount(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos totales por etiqueta: [855 744]\n",
            "Datos en y_train: [598 521]\n",
            "Datos en y_test: [257 223]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fhJ8hXLlRvF_"
      },
      "source": [
        "El objeto StandarScaler permite estandarizar los datos. Primero se construye el objeto, con el método *fit* se calcula la media y la desviación estandar del conjunto de datos, con el método *transform* realiza la centralización y estandarización de los datos. Si se desea tambien esta disponible el método *fit_transform* el cual primero ejecuta la rutina *fit* seguida por la rutina *transform*\n",
        "\n",
        "Documentación de la clase:\n",
        "\n",
        "https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "46StS2uUiw8n",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e4bd3564-71dd-4702-9a45-47e2cdd495e2"
      },
      "source": [
        "#from sklearn.preprocessing import StandardScaler\n",
        "sc = StandardScaler()\n",
        "sc.fit(X_train)\n",
        "X_train_std = sc.transform(X_train)\n",
        "X_test_std = sc.fit_transform(X_test)\n",
        "print(X_train_std)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[ 4.26040881  0.69402716  1.13621356 ... -2.64121515  0.54063364\n",
            "   0.64288463]\n",
            " [-0.60667593 -0.17731226 -0.90207707 ... -0.04913265 -1.15430921\n",
            "   0.07787101]\n",
            " [-0.13755933  0.83456577  1.44979674 ... -0.93589771 -0.21267429\n",
            "  -0.95798731]\n",
            " ...\n",
            " [ 0.33155728 -0.2335277   1.34526901 ... -0.66304693 -0.40100128\n",
            "   1.58457401]\n",
            " [-1.1344321   1.53725885 -1.42471569 ...  1.2469086  -0.8404309\n",
            "   1.11372932]\n",
            " [ 1.03523218 -1.35783663  0.97942198 ... -1.82266278  0.85451195\n",
            "  -0.20463581]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UBy3yzdKwDgf"
      },
      "source": [
        "Con los datos centralizados y estandarizados se realiza la llamada a la libreria **sklearn** en la clase **linear_model** y la función **Perceptron**, esta función crea una estructrua de clasificador a partir de los datos de prueba, el número de iteraciones y la razon de aprendizaje. \n",
        "\n",
        "Documentación de la libreria: https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Perceptron.html#sklearn.linear_model.Perceptron"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "viTTAoNyiw8n",
        "outputId": "b6163052-ec62-4bf0-cacf-92b7dbbb4de5"
      },
      "source": [
        "from sklearn.linear_model import Perceptron\n",
        "\n",
        "ppn = Perceptron(max_iter=40, eta0=0.1, random_state=1)\n",
        "ppn.fit(X_train_std, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Perceptron(alpha=0.0001, class_weight=None, early_stopping=False, eta0=0.1,\n",
              "           fit_intercept=True, max_iter=40, n_iter_no_change=5, n_jobs=None,\n",
              "           penalty=None, random_state=1, shuffle=True, tol=0.001,\n",
              "           validation_fraction=0.1, verbose=0, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4NZbUWZlv60g"
      },
      "source": [
        "Con el clasificador creado se procede a realizar una prueba con los datos separados previamente. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z2ofcSxviw8o",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "6a7232e2-4548-480e-9c53-b859c2095301"
      },
      "source": [
        "y_pred = ppn.predict(X_test_std)\n",
        "print('Datos mal clasificados: %d' % (y_test != y_pred).sum())\n",
        "print('Precisión: %.2f' % ppn.score(X_test_std, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos mal clasificados: 199\n",
            "Precisión: 0.59\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dJJlcFlXvkGr"
      },
      "source": [
        "Como se puede observar la precision del clasificador empleado es del baja, esto al realizar la primera prueba con los datos de Test. \n",
        "\n",
        "A continuación, se realizará el mismo procedimiento pero primero se ejecutará la transformación a dos componentes principales, para eso se emplea de la misma libreria la clase **decomposition** en el metodo *PCA*\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qwfmwekk3A5z"
      },
      "source": [
        "from sklearn import decomposition\n",
        "pca = decomposition.PCA(n_components=2)\n",
        "pca.fit(x)\n",
        "X_pca = pca.transform(x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "U8eNN4gy3ysU",
        "outputId": "c411a1dc-fd6d-4cac-a428-82804e498f02"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "colors = ('red', 'blue')\n",
        "k = 0\n",
        "for point in X_pca:\n",
        "  plt.scatter(point[0],point[1], c=colors[y_binary[k]])\n",
        "  k +=1\n",
        "#plt.scatter(X_pca[:,0],X_pca[:,1])\n",
        "plt.xlabel('Primer componente principal')\n",
        "plt.ylabel('Segundo componente principal')\n",
        "plt.title('PCA 2 componentes')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO2de5gcZZnof+90EsxMUEiDHi6ZHlZcFT0ui1lXVryscV1lVXZXz67DcAku5NDR6Dme3bNiFi/rxqO4imswwfEEufTorpdV0eMNEDReMSgXEVBkZhIRCQhCEgJMpt/zR1X1VHfXV1XdXX2d9/c83zNdt6++6k7et77vvYmqYhiGYRguhro9AMMwDKO3MUVhGIZhxGKKwjAMw4jFFIVhGIYRiykKwzAMIxZTFIZhGEYspigMwzCMWExRGB1HRGZEZL+I7BWRe0XkUhFZETr+5yLybRHZIyL3ici3ROQ1NX28RERURP4x4V7PF5GrROQBv6/PiMgR7Xq2fkZErhORs7s9DqP3MEVhdItXq+oK4ARgNfBPACLyOuAzwOXA0cBTgHcAr665/kzgAeCMhPscCkwCY0AB2AN8IpMnMIzFgqpas9bRBswALwttfwD4MiDATuAfEq4fwRP4rwceB1Y3cO8TgD0xx1fiKZJfAw8CXwgdOwe4E09BXQkcGTqmwHrgF/7Y3gM8Ffge8DDwaWCZf+5LgF8Bbwfu97+PiVBfT8JTlPcBs3hKdMg/thb4DvCv/vimgVfWXLsNuAe4G/gXIJd0LbAJmAceBfYCF/n7nwFc5T/zHcDfhO51MvAz/3nvBv6+2/+2rLWndX0A1hZfCysKYBVwqy9Yn+EL3GMSrj/dF4Q54EvA5gbu/T+AH8Qc/3/Af+DNRJYCL/b3v9QX6icABwGbgW+HrlPgi8ATgWcBjwHXAL/nC++fAWf6574EOAB8yO/rxcA+4On+8cv9vg7Gmwn9HPg7/9haYA5PaeWAIp5SE//454GP4SnTJwPXA/895bXXAWeHnmkE2AWcBSwB/tD/Do7zj98DvND/fChwQrf/bVlrT+v6AKwtvuYrir3A7/DemLcAy4EX+AL3CQnXXw182P88jvfmvTTFfZ+D92b8QsfxI4AycGjEsW3ABaHtFb7QHfO3FXhB6PgNwD+Gtj8YGnOgKEZCxz8NnO8L8McDYewf++/Adf7ntcCdoWPD/r3/C94y3WPA8tDxceDapGv97VpF8bfA9prv4WPAO/3PO/2xPbHb/6astbeZjcLoFn+pqoeoakFV16vqfuC3/jGnsVlEVgF/Ckz5u74IPAH4i7ibicixwFeBt6jqdsdpq4AHVPXBiGNH4ik1AFR1rz/eo0Ln3Bv6vD9ie0Vo+0FV3RfanvXvcRjeTGa25lj4Pr8JjeMR/+MKPBvMUuAeEfmdiPwOT7A/OcW1URSAPw768vubwFNKAK/FW36a9R0OTnT0Y/Q5piiMXuIOvKWO18acczrev9svichvgLvwFMWZrgtEpIA3C3mPql4R0/cuYKWIHBJx7Nd4gjPocwTI463NN8Ohfh8Bo/497sebqRRqjqW5zy68GcVhvhI+RFWfqKrPSjmm2lTSu4Bvhfo6RFVXqGoRQFV/pKqn4CmiL+DNiowBxBSF0TOoqgJvBc4XkbNE5IkiMiQiJ4nIpH/amcC7geND7bXAySKSr+1TRI4CvolnnL044f734M06tojIoSKyVERe5B/+FHCWiBwvIgcB7wV+qKozLTzyu0VkmYi8EHgV8BlVnccTuJtE5GBfyb0VKCV15o//G8AHQ9/dU0XkxSnHcy+eTSXgy8Dvi8jp/nexVET+SESe6Y97QkSepKpzeAb7cvpHN/oJUxRGT6Gqn8VbG38D3hv2vXieO18UkefjvWl/VFV/E2pX4nkjjUd0eTae8HuXH7exV0T2xgzhdLw3+tuB3XjGb1T1ajwbwufwjLhPxfO6apbf4Hke/RpvGe1cVb3dP7YBz7h9F56X0ieBS1L2ewawDM94/iDwWWKW8mr4N+B1IvKgiHxEVfcAL8d7zl/7Y34/ngEevO9qRkQeBs7FW5YyBpDA28EwjA4hIi8BSqp6dLfHYhhpsBmFYRiGEYspCsMwDCMWW3oyDMMwYrEZhWEYhhHLkm4PIAsOO+wwHRsb6/YwDMMw+oobbrjhflU9POm8gVAUY2Nj7Nixo9vDMAzD6CtEZDb5LFt6MgzDMBIwRWEYhmHEYorCMAzDiMUUhWEYhhGLKQrDMAwjFlMUxqJlagrGxmBoyPs7NRV93vr1sGQJiHh/16/v5CgNo/sMhHusYTTK1BSsWweP+KV7Zme9bYCJUA7U9eth69aF7fn5he0tWzozVsPoNgORwmP16tVqcRRGI4yNecqhlkIBZmYWtpcs8ZRDLbkcHDjQrtEZRmcQkRtUdXXSebb0ZCxKdu5Mtz9KScTtN4xBxBSFsSgZHU23P5eLPs+13zAGEVMUxqJk0yYYHq7eNzzs7Q8T2C1qce03jEHEFIWxKJmYgMlJzyYh4v2dnKw2ZINnsC4WF2YQuZy3bYZsYzFhxmzDMIxFihmzDcMwjEwwRWEYhmHEYorCMAzDiMUUhWEYhhFL1xSFiDxBRK4XkZtE5FYRebe//xgR+aGI3Cki/yEiy7o1RsPoedImrDKMFujmjOIx4KWq+gfA8cArROT5wPuBC1X1WOBB4O+6OMbFSZbCxwRZ+wgSVs3OgupCwir7jo2M6ZqiUI+9/uZSvynwUuCz/v7LgL/swvAWL40Kn1pFsH79wvZhh8FZZ5kgaxcbNy5kNQx45BFvv2FkSFfjKEQkB9wAHAt8FPgA8AN/NoGIrAK+qqrPjrh2HbAOYHR09LmzURnejMZJky1vasoTRrOzXrRao/+GajPvGc0xNBT93YtAudz58Rh9R1/EUajqvKoeDxwNPA94RgPXTqrqalVdffjhh7dtjANP7YzApXCDbHnhGQc0riTCfRmtkTZhlWG0SE94Panq74BrgROBQ0QkqJNxNHB31wY26EQtM4lEnxsIn6jljkYxQZYNaRNWGUaLdNPr6XAROcT/vBz4M+A2PIXxOv+0M4EvdmeEi4Aooa9aryzCwqfV2YAJsuxIm7DKMFqkmzOKI4BrReRm4EfAVar6ZeAfgbeKyJ1AHtjWxTEONi6hr+oWPo3OBpYtg3y+eUFmdUjjmZjw7D3lsvfXlITRBrpWClVVbwb+MGL/XXj2CqPdjI6mK/MWZtOm6hqisGDQLhTg5JNZf/kfM7lvgnly5A4o6/5mqLlsq1aH1DB6gp6wURhdopk17qjljiuu8BTFzAzr2cLWfWcyzxJAmC8PsXVrkxOBycnG9huG0RYszfhiJ3B13bnTm2Fs2tTS8kWmNaZdhnVoztvKMIwq0rrHdm3pyegRJiYyXdfOtMZ0LufWOoZhdAxbejIyJdMa01aH1DB6AlMURqZkKtutDqlh9AS29GRkSiDDJye9VaNczlMSTcv2LVtMMRhGl3EqChH5El6SvkhU9TVtGZHR95hsN4zBIm5G8a8dG4VhGIbRszhtFKr6rbjWyUEaLWD1IAzDaJFEY7aIPE1EPisiPxORu4LWicEZLZJFYZseUTQ9MgzDWJSk8Xr6BLAVOAD8KXA5UGrnoIyMaLWwTUYV1FpN12SF3Ayju6RRFMtV9Rq8KO5ZVX0X8BftHZaRCa6kf6H9sW/qGVRQC9I1BXFzQbqmRpSFFXIzjO6SRlE8JiJDwC9E5E0i8lfAijaPy8iChMI2iW/qKRRNElmka8pgGIZhtEAaRfEWYBh4M/Bc4HS8OhFGrxOV9A8qBYo2nn1v/Jt6BhXUskjpYYXcDKO7JCoKVf2Rqu4FHgberKp/rao/aP/QjJYJMr3m85GHdz4aXUK28qaeQQW1LFJ6WCE3w+guabyeVovILcDNwC0icpOIPLf9QzMyYWKCqT2vZoxphphnjGmmGAdglOi1m8qbegYV1LJI6WGF3Ayjy6hqbMNTEC8MbZ8E3Jx0XSfbc5/7XF3UFIuquZwqeH+LRdVSSbVQ0DLoLKt0nJJ6lgjVYfZqiXEtMa7D7K3sB9XhYe/Sdg8vjD9UFfH+Zn1/wzCiAXZoChmbRlH8JGLfj9N03qm2qBVFsahVkj5oS5ZUbe9luEpZFJhWBS0x3lUhXSp5yqndysowjHrSKorEwkUi8mFgOfApvNxPfws8ih9Loao/bsNEpyEWdeEiV6WgCGYocAwzAAhlyuRgzRq4+uo2DjCesbHGq7EahpENaQsXpfF6+gPg94F3Au8CnolX6/qDWD6o7tOA+1DYJjHKzq4rCRgc11eLHDcGmTReT38a017aiUEaMTTgPrQTz0o9zD42yflw1lmt3TsD6TgIrq8WOW4MOk5FISKn+X/fGtVavbGIrBKRa/0cUreKyFv8/StF5CoR+YX/99BW7zXQuNyHllQnBn6E5WzkXygwwyTnMKGlutDmhuR+RtJxEFxfLXLcGHTiZhQj/t+DHa1VDgD/S1WPA54PvFFEjgPeBlyjqk8DrvG3DQdTL9jCZSNFDpBDgfKQXwXu0ksX/EmBYfYzxenMcAwTfMq7uCaVR63cP/10eMfLvhOtPTKSjoPg+jooy2eG4SSNxbsTDfgi8GfAHcAR/r4jgDuSrl2sXk9RHkOgms+HvIZKpQXf1NpWKFT6KhSiT1nuu9LWuSSJRF8gkmrcG/Ilnaag84juyRcyd3PqpMut67sLfb2G0ZOQoXvsZcAhoe1DgUvSdJ62AWPATuCJwO9C+yW8XXPNOmAHsGN0dLQtX2Kv4xJQgTzfXnRokggfVJfcD7vSVknAfD765Hw+dsylkurapSXdS/W45pZl5xPbaZdbc/E1+pUsFUVUHEXdvmYbXoLBG4C/9rd/V3P8waQ+FuuMIk64g+quXCH6QC5XJ8XilI4wX7NDmlYUhYLqNI6bZfQK3o03fAsaNPqRtIoijXvsUNigLCIriS+hmhoRWQp8DphS1f/0d98rIkf4x48Admdxr0FjasozG8Rx5LxjkbxcrjMCbNpUMWfUUZfqY3QUHngg+mTXfp+dO92pQ7Ja1O+GzWBiwov7KJe9v/1kYzGMJNIoig8C3xeR94jIvwDfAy5o9cYiIsA24DZV/VDo0JUsZKc9E892YYQIDM9JIRSBO2wdtb6nU1NMbBzjCp1gOdUG6mH2sYm3h3b4LklN+rWOjjYwriYZBJdbw+gp0kw7gGcBb/LbcWmuSdHnSXiR3jcDN/rtZCCP5+30C+BqYGVSXwO79ORYz3AtreSY06Xsr2yPU28LqFs8r1lgLzGuBZlVoezdsrg9ek2lyYX5QbRRGEa/QlY2Cq8vcsCRwGjQ0lzXqTaQiiJG2rltE2W/LSiK3eS1DFoO7Ae10rKVBf0mF+Zb9XpKc1uzGRhGMpkpCmADcD9wq//2fwuWPbb9xAjwOMNzQ7MJ1Vg3114UtjZbMIzsyFJR3Ank03TWrTaQiiJOgBe316UHX85ePZHtle00nkWlkmoht0uFeS0wXRUvUcpv6EmBbDELhpEdaRVFGmP2LuChFswgRlrCOTRcLk2jo3z38l/yKE/AM/EoIzzMxzmHT3LawmnsZD2bWcIcQpklzLGezRXXn0ok9vzRKEPMMsY6Pu4VNRoeZiPv7cm0FBYFbRidJ42iuAu4TkTOyzLXk1FDbQ6NKJem4WHWH/s1tu47w0sRjgDCPg7mu/xJldvpGXyCrbyReZYAwjxL2MobWT/8CcCRgYMRNuYugMlJdj6wInKYzQrkrLKrmkeTYXSeNIpiJ3AVsIxscz0ZYaIkN3jZYUNJkCavewaegggjTHJuldvpJzk9+rz9pwMxb+blo2FiIjuBPDXF1GFvZt1p+zLJrjoISQQNo+9Isz7V620gbBQpcyeFvZpqPZ42cKEK83oo98caulWT1/ozMRr7nRSYztSu0ItGdsPoR2jVRuFXtkNEviQiV9a2jmmyxUKaV/j168kRHWWXY56P8D8pk+O3HMaQ6zy/fEXUm7mI97Y/NuZtp8rqun69l9JcxPu7fv3CMX+W5Aqwa3YZy6KgDaOzxC09XeH//Ve86OzaZmRJkuSemoLJSdZxMZ4RO4z6+/3LoGo7TFC+ovDdKW5/dIx5hphmjHGmUL/bYGkIEgTy+vWwdeuCPWV+3tsOlIWvCVwpO8yuYBh9Qtx0Ay/QbirN1KSbbSCWnlQX1lSCJSfH2lGRzZpjTqGsOea0yOa6c8qgIyMLu4aGVItF7zbbi/UxFntZru/g/MaWhlzpy3M577j/LCXG69x5e8HV1jAWO2QYR/EdYFmazrrVBkZRBKSJqEtqgbCOWNB3ZZXdxVE6DzpNQb/GGp0jt9BXoGXCJBlCQoaOEuNaYNqL2cjvMSVhGD1AloricuBHwPnAW4OWpvNOtb5WFFGW2aT84WlbPq+6bFn1PhEvnUdEm2fhvpHn1CqLpBmF6/kWKfZVGL1GWkWRxj32l8CX8ewZ5h6bJa6607W2iho01GL57W/h8cdrLlbmyUWe/muOrHwWYIZRxpjmZXzNC9zb+tFqe7WrXnd4v1megcxKjBtGVxBPqaQ4UeSJeEsKe9o7pMZZvXq17tixo9vDaJyxMU9i1DI05AnWNlEG9rOcEfZX9u1jOT/heE7i+6HzhBxlPJVUX6yiWIQtrPfcoebnPZeqdetgy5a2jb1fcf3UhYKnPw2jG4jIDaq6Oum8xAJEIrIa+AT+LEJEHgLeoKo3tDzKxY7LP7SNSsJD+AnHM8ZOjuTX/JojmWG0SklAuG5EdEWjrVsV1ryZLQdMMSRhqUeMfibN0tMlwHpVHVPVMeCNeIrDaJUu+IcqMIRyEt/naO5mCOVo7q5TEvsY5u0khTsLk9cc29z6SVY5PfoESz1i9DNpFMW8qm4PNlT1O8CB9g1pEREVO+FieNhdqzQrcjnKCDMUOIdJPkWyPWGeXOOZAju4YN8r+ihNmIxh9CxJ1m7gw8DHgJcALwa2AB8CTgBOSGMxb3fre6+nOM+lsIvMkUe27Anl8ngKPK2iYh7caUO8qnq1aUYS6VCu8F6rXREXJmNxJUY3IEP32Gtj2jfT3KTdra8VhWq8cA9YsyYTJXEVa+qC7cpUS62qmAemdQ1fU2E+QmGUvWC/RgV8yrxWrdJOfdSKq6vV1DB6hcwURT+0flUUxaIXcjDNqFvABxIotG8Pw3oOWxXKkUWHXErituP+MroMaZyCCUnD4nHX1keEL1vW+KtwhyRlu/RRqzOVDulJw0jEFEWPUywuCIjNFKuC3eqEZ8Ty1F6GdZzSgqBib0VZ7GO5nsoVFQXyCAfpbvLu+tQOwb0nX6h+ay5uV126tPq8pUsbVxQRknYvw7ohX2p5+SX8pu+KB2xVH7Wq52xGYfQKpih6nECIRda2rn3NzOcjj01TqNo1yrROU6hSIMvZpxt5d7xQrhHcJcZ1lFmNWmoa4kAlt1RLaTlKJd2T92Y14TG3slYf9aZf27KwBbQ6I+g124mxeOkLRYHnersb+Glo30q8Qkm/8P8emtRPPyqKQEA4a1sHLZzZr6bNI7GXbqaoc+QiDdjTFKqFk/8qXuJUHZZ9sf1CWfPcp0vZXy3sls11da3e1V8ul23ajCzGbek8jF4gS2P2MF6ep4/7208DXpWm8xR9v8j3ngoriguAt/mf3wa8P6mfflQUwYzCueQUarXG5WCJKXgTD+wN4TfzzRTdHk6gV3BqReiHZwMuIbiS3RHeUBHCkunUki+LtfrAzhM3pqzX/m1GYAwKWSqK/wD+dyDMfcVxY5rOUw0AxmoUxR3AEf7nI4A7kvroO0VRKmlx5FKFcuKMIjJFN3t1G2t1M8WIdOGe7aKS+dXR53KqZw3BbMCdj9DtIlsllJmvkpxxb86tvpmH7Tyxyitlf41gMwJjEMhSUezw//4ktO+mNJ2nGkC9ovhd6LOEt2uuWwfsAHaMjo624SvMmAgn+iKbdYLLY20UrjKiw+xxKpmy31wzkVXMOmYDM1pY4Sqjmk5RFJheUEj5DbFv3q2+mSfNJOxN3zDiyVJRfA9YDvzY334qcH2azlMNIEZR+NsPJvXR8zOKFFZW1zKRF78QfVncspVrJlJiPEbol5WIeIlh9mqe3clCOeR5Fafkwm/4rbyZx43F3vQNI5ksFcXLgW8B9wFTwAzwp2k6TzWAQV96KpXSvfo2OKMYZTp2ecl1XZ77dDRhlrKGr2ttjEaU4lnKY5pntzOWI07JZbFs41omC5fDMAzDTaZeT0Ae+AvgVcBhaa5J2yIUxQdqjNkXJPXRs4oijb9mwowibmYQd51bSHvR1LXHc8zpNtZWdhTZrBdxblX/ntvsdNWMQ1WdEtulrKR2xpK0PBS2WPvV9kol1SVLor/OqGJ8hmHUk+WM4po0+5ppwKeAe4A54FfA3/lK6RrfPfZqYGVSP72oKEol1UJulwrzuoujYpVE7RJStW1hRotsjn1zD2wS6YT0fGQupz/hW5H2jCiFVpkVMBNrUW4kb5TT4Ozo/9KRYmQ/K1Z08lc2jP6mZUUBPAEvpuEm4FD/80p/BnB7ms471XpGUfgL7mVEZyUURMaeSAFcBj3AkM6nnEHEubumE9KqxMw0ou5Ze+IcufrjjhlFiXHftlFWl/2josBcLqyOZbtgHKn7MQyjjiwUxVuAaeAx4C7/87SvON6UpvNOtZ5QFI60FOEoaVBdy7Y620J4RuGaCWzgQqdSiFu2WlUTYR2Ou7iQDX7uJlfX5SrlVga9iHNT5ZaKUlQ55nQl90Ve4pxRxDxzQ/0YhlFHlktPG9J01M3WE4rCERRQm2YjzqVVcdsWXIkDy6C7iU7xMUdO38H5VUoi7IrrUkq1bZi9WpKJ+JNq7AguN9t8RODesOxz2ygcM4r5oZwFvRlGi2RtzP4T4FTgjKClua5TrScUhWP5ZSEC2mtXcGqsonAJb5cr7DzE5osKz2pmWVV1LEopuSK9gxlBkOepqkW4GTmjrpmPiO841f29umwgIyNaKm63oDfDaIEsZxRX+LEUW4DNfvtIms471bqqKMKBdDUtWH6pFr6juo0z6s7d4wt6l20hbiYyTaGS1ynqnGlGtcB0osE7SuHUL5+V65VFhJuRM+o6FJC3sLMQ/x0Xi6pDQ1Xfa10ywgjPKMMw4slSUdwGSJrOutW6pigS3F/z7HYK3+2cWNnez1K9n0Mr29s5sSIIj2KXU4hXzyykyiheeyxOkSUpo9rlsxxzVQK5VFIt5EMG+/wGLRW31301By2Z00tCLrgK6deL/Ay6ZdBZVlUpr6U8Gm0zMWVhGLFkqSg+EwTA9WrrmqKIeG0O3uqDCGiX8N3FUTqP6G7y+ijLIs+pfZv3ZiajTuO1a/8ujnKm8wjvdy9v1WepDSiVvDxRVbKfvVpaujZ6aShlKHb4tA35+txVtd9NMFOpdS225SjDcJOlorgWeBD4OnBl0NJ03qnWNUURsRAfCNtgWSdJ+LoUyRy5Oo+psECMm1lUC9Tluo0zY4P2klrtjAIW5HvsElODLkiumtJpZjqB7aPuOTts4LZkgUY/kaWieHFUS9N5p1ovzSiC4LpTuUKnKTjf8gMh51IkZXAK9z0xS1Bl0BlGdR7RGd8e4sy5lKB0FPRRlumJbK87FAjgOKN1YlBDyK4wP5TTi4eig+jSzHQKTKfKLdVOLP240W9k7fVUAF7mfx4GDk5zXadaV20UNZJynuhI6eq3/IVlk7g040HKjNrlog1c6BSes6yqUy6uILtKSvCYto0zNC6auukZRYQ3Uxl0M/XKImlGEdgoXK7FnQrCsxKnRr+RVlEMkYCInAN8FviYv+so4AtJ1y0aPOVZYQgvN7pEneq3x1lS2fd2NrGP4cjzJvgUsxxDmRwzHMMEnwLgIt7MFs6lXHOXMsLR7GKaMcaZAuARRlwD5zQu5wEOZi/DaN1RjzdwOVJ31GPnTti0CYaXHajaP8w+Ni19t3fQxeRk3S4BzqV+f9R39AjDvB2v/zkO4kEOYRW7Im81OuoeRpbs3NnYfsPoG5I0CXAjsIzqehS3pNFCnWpdmVE0kPAvqj3KssqsYpyS7iafOkVHkPJ7nJLOsiryumpjb/SMIMdc3bVeWhHR2zi2sn00O2PflGu9nraNbNAN+VL8Or3j2VwR1+OUdOdQoWrxv9YeUCx2d+nHZhRGv0GGNoof+n9/4v9dAtycpvNOta4oCpdUaKCFjbGJtbP9VmJcl/Foze6y5tldZ5xe6N9df8I1rrCx+0I2RF5fJ4BLJd2Trw/WixLWB6SxHE6QbgmpEWNy1oZns1EY/UaWiuIC4O3A7cCfAZ8HNqXpvFOtY4oiLFlaVBJKtTE2Te1sxbNPuA7XejKV/Xu48jlV4iHqxkUlQDDYrj0tn4/4bhy5rsYp6a5coUoiXxRR09tlowhalm/m7RLq5vVk9BNZKooh4By8eIrP+p97KgCvI4qixaWmqPYbnlwxVielIq8Vvk5hGuHJVGSz1s8qIiKsQwqsVoA/xEjsDME1w9pNvj5QcNh7hiCavIw3kwiURCfezG2ZyDAyVBT90DqiKFpYaoqyIZRBv82f6DzectIGLoyNvA63YEkpKi+Ty5NpG2fUuc1GjS1uZvMoy3RDvhQtsCNmWUU2+7OZcl2eqKjYjKCVSu1/M3fd29KUG4uJLGcULwCuAn7OQrrxu9J03qnWEUXRwHLTPg6quMjuJq+bKVaM1YFgTuNG62rzSCWlRzgSeRWz+nK+kqrPMqgedJDetqaoO4cWlI0rDUjiK3eNIk2axURFewet3ev6cfEfgVKy5SNjMZClorgdeCXwZLzqc3kgn6bzTrVuzCjCuZmi9gtzkW/8wdLQAUQPMBTZR1KbYVSnKURGIi9nn17IhvT9+QTCMdGo7nrlrlkvSrKLxM0o4vRRG37KqkfrtueUYXSSzL2eerm1VVE4ssPuZmWktNnNSi0xruOUdBtrq/IrbWOtjnNFaiEeNTPYx3Idp6TziDMSeZg9zvxRLkURPOqGfCl2+WlPvpD8XYlonKdVbTGlRvRRFsRNDs12YSwmslQU7wM+AJwInBC0NJ13qrVNUcQYsOPTSpT1QjZEpuC4nImqzLGuNkcuMhfTqd5tBqAAACAASURBVFyh4L35uyKRwbNfzCXMWPYwXFliyedVly71Dm2mGLkEVVkqS7EeM+S8dbyC6IRgjlMGzpQkTSouW8YyepksFcW1Ee2baTrvVGuboogxYLsqzgVLKqMx+ZX2sTxWSs4jKszraETJ0WAWMU7JfzN3dxU3M9jPQbpWLo1VNGG7St3sJmE9ZmQkSim4ZxkNdN0yUfo/UBCOgnoNKS5XcsNOPJthNIJ5PWVBzBrFej4cW+THmXcoRX6l8Pp9bWxE2C5xItt1OfuqBZGjvsR2TqykNt/FUXrOkm1Jw4jsJ630FFGd4NIqrydXziko6zB7FLyZTSPpyJslTpg3qBPr+k3yorZlLKNXyHJG8STgQ8AOv30QeFKazltpwCuAO4A7gbfFnds2RZGPrkUdVhbTvsvpNKM6Tqki8IK/dUIiFOfwGEN1b+pRcRK1sRFhT6c8uzXP7oodJBxFHfaMqq+al+7tHuKztwZv4LWyvFCovy7OphIWytuLjQVStKJTkryes+4vUEyG0QtkqSg+B7wb+D2/vRP4zzSdN9uAHPBL/37LgJuA41znd0pRlBjXS5moBInNI/oQI3ost4WE7IKgrfX8Cc8O5sjpdk6slDINPKM2U4yMjUjj8rqPYT2VkhYKC0Mfp1SpkpemrWK2TlGlrXwXluWlkupMzXWuMq+1bVcu+n5Rr+KtBuclGbYbJY0Xtc0ojF4hS0VxY5p9WTbfcP710PZ5wHmu8zux9FRiXNeyrW65KfBCShIOOeYijdPhJHiusqkTXB6rKILI5qtYo/ev8JTMhWzQJQ4X1YVWYxvwFVmtAkhTSzscZR2USD1npP668GzIpcCctpWIV/FWvZTiZgDNvPknzSjMRmH0Elkqiu8DJ4W2XwB8P03nzTbgdcD/DW2fDlxUc866YDlsdHQ00y9v+3Fn69lsrayvCwd0iAOp36zrhIOjmlwZ9EaOq+yK67+2FKhLYYS3L+V036gevcyUZ3dkedSoYLiomJCwkohSZBdRdGfGHR7WDfloBXsUuyK/r1J+Q90SU6teSknBd40SZyg3ryej18hSURzvL/3MALPAT4DnpOm82ZZGUYRbJjMKf6F7OyfqOi5SKOs4pSo310ZqSnsKploAB8I8aL9gTJeHlmLi+o8SxmXQS5nQT3B6rAKJLBEaUwo1SfHVNpcSC2eC9Yz7nlH7dC6tpAl3GX7rjPhL19bX5h52m5GihLzLllEsZuudZC6xRr+QudcT8ETgiWnPb6V1fOmpVKoEEZTxooqD5ZawAdb1xr+bfMTu+TrFUCvkd5PXd3B+xZaRNGM5lPsrbrOXM6FfY03lLf9BnpSoLKJmD4+ytOq8pKSDUc2lcMrgjKcoFhe+eqdLam5XRdoW8tHOAfl8OhtFki3DhLuxGMlyRpEHPgL8GLgB+Ld2p/DAq3lxF3BMyJj9LNf5LSuKmtdSKOtR7Kq8BQeHotbqFfQrvDxCiLmzs7oEs8sWEFWz2jWWRls4VmI3+cpYcjnvq0ljnHXOTnI5p6II+o+7R3j5yF2bu+wtSQVKML8nUsgPXMS1aTYjA7JUFFcB5/tC+xjgn4Cr03TeSgNOxktE+EtgY9y5LSuK0Bt3redSlIAO12pQqAuuizonroXTbMfZAsItbaGjRpIO1s4mVNMnzQ278C7MXuKjsAPSCHHnOTJbvcOxZpR1xHVXsQpJRkZkqSh+GrFvoEqhxrlthgV3IMxr36DDwXWtvOlXF+4p60E84lRccfaM8PLSbvJ6P4f6NSVW1C011bZgmSuXU9ViUeeH6utFRApcyqq5XGoX2KD/pGt8B6po2Sj7omcyEdMEly2jrgBTh2lqYjBw0yOjW2SpKD4EvB6vgNEQ8DfAv6bpvFOtFUVRKtWn21hQDtFFfG7j2KpaC2FBnvZN39U8A3BZj2RGFS9TbFAjO9zSeGAtZ6+/bFWuxHoM49W2TjLMv5BrG6pAV8jvUdV0M5BxSnr/kqdU7UyahQTKokqocmr0DSKmCb2oKJqeGAzU9MjoJlkqij1AGTjgt7K/bw/wcJqbtLs1qyhKJdWLh4oaTi0RFStR26JrLXgt7XKTqwVxFcK8lhjXeUTPYFvdG3ea2IZqpVe9lLWbaMkZGObjPZki4i+WrlUtxRnBPYU6weWxmW3jamanraoX9WadhRtt1iaBpicGNqMwMiJzr6debk0pipJXx7nEeNXS0cOsSBTmrloLkNWMwv9/z3RlyefjrK2UTV3FrL6Qa2uUwGikknApkyRFEWfb2D7y8qq8UUE23FJ+gzsmodaW4GjhAMTaVvu2XSpu14LMVntyOV7Jm5Wta9akG0szNK28zEZhZESmigJ4DvAa4K+Dlua6TrWGFUWppMWhi/VMtlUJ/XFKqYy/ScbuKOFce+LjEXmeapd2JORiW3t91OxhlOmqWUNYadVGRLuXnrzlK9f3EDdjcuVyCmZHrmepvX+Kn0CHhupTmQ/LPi0Vt7t+8oZlq0tJuJSMb3JRWLCtxNHSxMC8nowMyHLp6RI/Avoy4BN+uyRN551qjSqK4silkcs5aWYDZdAhDsSeVrvcs4ELq06YI6en+gWM5hGnsbg2GWBtqw2MCxTLXob1WG6riixfwuNV5z7EiFNQxz37HnHPuNz1MbxxFdnsTM8eVkRpFEVa4R2mUdma5n6Vf1PF6ONxysImBka3yVJR/CxNR91sjSqKHHM6ykzdf+rgbTnqTT/Z88edjTXP7qo36nBJ1ECA1l6/lP2VrLBpI8KDlBxJLr6gDZVhDT/7FZwaGbinoCMHPeb8DkqMVwUyhk+o3U56k49rWdpz09wrEOquoMFwvEgtpVK1kb2SYt0wOkSWimJbXObWXmiNKoogvUbtf+pgRnEbx1YS3C0oh7Ku4WuR6bqHOKBr+JovpOcrb/JV6bN9l9rd5Ovca8OKIkgdvoxH68ZV26pnFPNV1yS1RuIrwgopqv5FiXHVfD6mqp1nowieMcrAXrnXUC6xnkNcc80omlmpaeR+cee4xmOzCaPbZKkoXgw8hFcb4mbgFuDmNJ13qjU8oxiaj5xRhN92w2v6w+zRER7SEuN1qTBezlci35A3cGHd0lCgQMIzjNrZxG7ydWv96Tyc0pUYDVqaJINBCwzcTkM106r5fMIY3LOcHHOq+LmrRtyxGknNJWgbFcphW0NSC2YwLiU5NBR9j7hSJ2ZyMDpFloriTt+QfQxQCFqazjvVGrZRFDXSRhEIZXdktDerKIPu9wPXhPnIa4IaEmXQy5nQpTVv+8GbeK0HVRAw19i4AkGcXqhezNmRS2xxisIpLH0bRCN1L8LjLrJZ58j5rsqNXT8ykjxLaMRo7LI1uNqKFd51I9EmHx0Zqb9HqZTcr80ujE6QpaJoa0rxLFoz7rHFouqZbNOV7HYIWbfN4Sw+XhHccbWxgw2XN1BUCvBpCs7znUKFvZFBedHPU46sxR2nKAJPKJeSCrya0kZlx7W4mtWNehUFNOKG2syYG013njotSqHhf9aG0RBZKootwCeB8YFxj62hdMh6PYTfarA84gldlwdPfT3sXI1HUVyKj6Q2Tkm3sTaFwK1OY+4ud+ql9j5XLlYtFnXP0MGRHe5luTO2otaYHleuNbw01+gsp/L9Zbx27xLMQ0PV/TY6mwgL9EZmLWkSLbqUjGFkSZaK4hMRra/dY11sX/pifZjlWsY9CwgLYCjrCA9pkc2VYLhaTyBN7Ku+BXUw4ooOFZjWMl7d7WCnK5V4leRx3HQ7J0baQlweVy7X3DTPnVR+NOsQgVAW+cgWzEzS2iWinqcRO4jNKIxewSKzW2DqEM8NNN0bsbfG/lsO1vN5RyU9eVhQp1uSqVY++LaPy5lwFh3aXKkgt7IhL6Y4wV9rC3HFVdS65gbPG1ZURTbraVxW1d9pUtJiMdqY20x9iLTnxRmPIdlusGZNciW82rEUi9FjiyvYlKRkDCNLspxRHA18Htjtt88BR6fpvFOtHTWzP33c+foCrkulLHLM6RWc6nYd9ZXFodzvvYEyrxNcWrXmXxtc90Y+EumBVWBat7E2NqdTIx5NSnyxojSuucPs1XO5qE6hncqlkd5a29Z4EtAl5F3lRGttEo28xSct9xQK7hlFsATUSCW8uLFFxU+4lIphtJOs61Gc5RcTWgKsBa5K03mnWjsUhaqqlkq6TiYjs8RWt3KswXqOnH6UddXCh/lKYaO9LK/kSwqaq2JdfQbX+jFFLSHNsSSyv10cFakkgjfiU1MmHwzHjCQpmRkKsV+7a2kmHOAWd14w/rTnhhVR1LGwgko7g3HdL21FPsPoBFkqihvT7Otma5uiCPClgysZYI65hgzWYWXhKiMa1wJBHXfP8AxjT76gxeOurSi8HHNaZLOezcUaZbQPC65CId41N+5NPS6iPI4kG0aa82qfI2lpKei3Wc+qRp4h6bkMo1NkqSiuAU4Dcn47DbgmTeedam1XFD7Rb5yejaJRg3VFQCTkc3JrGdHckFtRhN92XeOOmo0sW1b9dptG4LnOictQG/lm7ufuiEuXEvYEasQoXCq5k/zVPnMWpB1b1HMZRqfIUlEUgCuB+3wbxReA0TSdd6p1SlGoekI3HIX7Kj6vZdIarCMERIS3UNqWNhFdI948Oea0zIL0TivwwmvuwQzEZWAPAhFv5LjK7q+zJlWhpPCbdxqjcFgZRHk+Rdk+GiHOzrLMXXYjUpkZRqcxr6c2EyWkli5N9q6pExC+m2vDisKXLGmWShrrurywMTyst60p6qzE1/AOe/2cnktfCjasLFzfQbg+R9RafiCom/kK4/pt9t9AuC/Xv4W0BnHDaDdZziguAw4JbR86qHEUjdKI106kgGKvXsiGhjK5lhjXURYS7KXJONrojCK8IynYbtkybwzBd9Bo4aagUFHc7CONJ1Cjb/BRLTDgp83omhRkl2RrMQ8no9tkqSh+kmZfN1u3FEUcUUrEM6jW2wU2U0w1qygxXueCC95MxuWiuSDM0seEJJ24K1dQEU+I1i7nNFoKNklReP9E09HobC5tc9kwktJ2WMVSo9fJUlHcBBwa2l4J3JKm85g+/xtwq19/e3XNsfP8RIR3AH+epr9eVBQuimtuixDaZb2C1+v8UPyrf5zBvFb4RM9qogzYC2k+0iiJsCSMEoTNzii+FmOjqLzV12jf4prbqpbdYCGqvdZe5LJRpG1Rwj1JEVgqcaPXyVJRnAHcDrzHb7cDp6fpPKbPZwJPB64LKwrgOF8xHeRnq/0lkEvqr58UhaqnLMKuqs9edlu18HBIoCQX3PAMxvV2HRQ3ikrzUTsbcM0O9uQLqqp6EcW6uh1RMRxxSiJs0A6URZTX00FL5rS0dG3l2qhiT1DWs7lYleogxYOH9kQGujXSoryS0igCq1hq9DKZGrN9Af4mv2VWxChCUZwHnBfa/jpwYlI/PacoWnXGdxg54mYUqRPNObys9uIZrrWwYLjeTDEy2G5CSnpRxHJZMANYiLuIr5EdVhKp3upDrsRxMS0uxRb+ehs1gGdZEMkweoW+8HqKUBQXAaeFtrcBr3Ncuw6vlveO0dHRzL/ApmmmeHIUYQnk9xHlghtls0grbOcRncezOWwvLki48CO4gu1caULCXkrg1b2om5n4NBxrEFJycVHytTtLnNqS00E74iwMoxfouqIArgZ+GtFOCZ3TtKIIt56aUTRTPDmJKqFXnffJS16YTtBWpT/3Awii3ojTCPA4L6UgUj0yg23o1byROIhaJZd2RlFiXIelJgdXTPLBRryeDKPf6bqiSHXzQVx6ipN0zeIKKfYFb1xeIa/ymmewHuJAlcF6mkKkT38aod1o4sE6Ce0TpZjCsxjPML1HvToh9zlLyAazicBGETSvVrfzqzMlYCxqsrZRPAV4ld+enOaalP3WKopn1Riz7+o7Y3Y7ZhSqqscd5xS8LqNqtH4pV+p/16YKDw81SeZHufSWwbtplLZJeDUPlrtq65a7UqxrPq/FNbfVKboT2a6zUqhEl0uCW7B5IRmLmSy9nv4GmMULvLscmE6zHJTQ518BvwIeA+4Fvh46ttH3droDeGWa/npKUWRlo4gixnIadShO4AfBfnECNI2yCHs9XTpSTBxn0lc3E3KvdRnvg6Wt/Mj+uiC72pQcaZbRLK7BWKxkqShuCs8igMOBm9J03qnWU4pCNbsUpC2SuBIUkRo8vCQTCNk0HlXhN/O0jx+pT0I3ayYjb63gT2sHMc8lYzGSpaK4pWZ7qHZft1vPKYoeoZnSnnH5lAJBur1Y0j35BU+oDflSXabaE9leqfZ3FLv07OO2V3VWRnRWqnNHDQ+r58qaMKNIarUxD0kGesu9ZCxWslQUH/CNymv99lXg/Wk671QbWEXRopO+axXMJShT3SIhykxEdSn7tdbQLMzrs3O3VgXNKfW5ozbkF/pvNiNvXMxDVOW8RvpIwuIqjH4ia2P2a4EP+e2v0lzTyTaQiiKj/A+1adGbUg5hYvJWJBUHEuZ1Kfu1xHiVm+9R7KqeDYSkbSm/QY9ccq8K8zrKtF7IBmfZ1uD6pOSBYUGedlaSBkvZYfQbfeEem1UbSEXhyjXR5KtuZmaTmEx4aQPohjhQN+MIWi5XLVi3F6NLscYlOWyELBP3WRJAo99IqyiGcCAie0TkYVdzXWdkwNQU/Pa30cd27myqu8sug/l5b3t+3tuemmpibKOjzv1ph1YmB0jksfl5WLduYWw/ufiHPItbGWKeMaaZYpwRHuEo7o68fmQk3RgCNm2C4eHqfcPD3v5GmZ2t3zfOFNfNjsHQEIyNNfmlG0aXSdIkeIkA1wMHA08EisA/p9FCnWoDN6OIezVv4vU00zfdmPWVLNN8B7MeVxzFFZwauaQGrWVLadauUCpFTbbqAwBtLcroJcjSPTbNvm62gVMUcf6oTQiZpLoJDeOQrFnXg3AampnWXbmCU1GEYxs7ZVx26/aINO62FmX0CGkVhXPpKcQ+EZkQkZyIDInIBLCvDZMbI8C1vJPPw8REZt259icyMQEzM1Aue3/9MT3wQJP9OfDeSerZySgz6zZRLkcfD5bYpqa8ZazZWa+v2dnqZa0scS+7CSDMs4StvJH1bG5q+dAwukkaRXEqXnT2vX77b/4+o124Fs7/7d8y7a6Zdfg40igeiTZNNMSTVzzCSVsmyOXc50xNwcaN8Mgj1fsfecTb3zDr18OSJd4DLFnibYdIp3SFSc6FlSubGIBhdJE0045ebwO39KSa+ZpJJ5ZgGomCbnb5KbzEH5crMS7YMO2SW+U7oxydCTdkDIm2UUQvRTlr1xpGhyFDG8XhwNuBSeCSoKXpvFNtIBVFH5KmglxcmdDaNjzsyeJma0m0EkwXabMPp2mvNYaoN9YkZVFJg252CqMHSKso0iw9fRF4El59if8XaoZRIbAHuLx6A2ZnPS9RgMlJKBS81ZxCAYrF6u3JSdiyJdIcErmsVItq/VJX2iW3yGUrRtjIexd2BMYQny1b4IorvLE7RsQ6LvY+mp3C6CPEUyoxJ4jcqKrHd2g8TbF69WrdsWNHt4exqBkbi44jcDE87CmCJmzzgBeWkPBPt0Kh4Mnl0VFPSaS5p6t/oezHgQC5HBw44Oxj/XqY3HqAeXLkmGcdF7OFDQuDmplJ9wCG0SZE5AZVXZ10XpoZxZdF5OQMxmT0KwmGXHC/IAezg1qaNir7pPXYCuRx7YwEvFnQ2Fh0LJzTU4zQg65bF3vvLVvgQOk/0OEVHGDpgpJohyeBYbSTpLUpYA9QBvYDD/vbD6dZ1+pUMxtFG0lZXyMuqC/zOA5Nb+Nw2YyT8jLF2igazYFimQKNHgXL9WRkQsqKfXGCt105kKLqXaeVx2nGZPLdGHTSKoolSTMOEXmRYyby7cymNUbvUmOwde0PG5mj7AHr1lUbhxNXX4JAiFBnU0zU9d/sMr9rqSy8f2KieRuKYQwSiYoC+IfQ5ycAzwNuAF7alhEZvUUuF60sIqLdXII1SYnUEbhQBZpldpYDb1jH1QqzcxPBroqJoBlhPjoabXxvOlrdMAaYRGO2qr461P4MeDbwYPuHZvQELoNtgiG3FkfWj2gifFOXPP4I75yrtn63YhDvVLS6YQwCabyeavkV8MysB2L0KFu2eAEOuZBLaLHo7W8XO3cyxThjTFelF6/yOFo4tSkmJurjOFpx1zWMQSZNHMVmIDhpCDgemFHV09o8ttRYHMVgMXXYm1n32//DKXyB97KRUXbyK47mVp7JyXy96lwLRzCM5skyjmIHnk3iBuD7wD+2qiRE5AMicruI3CwinxeRQ0LHzhORO0XkDhH581buY/QnG3kvp/AFPs46xphlCGWUXbyI7YyzEOxgS0WG0RnS2CguC7UpVf1uBve9Cni2qj4H+DlwHoCIHAe8HngW8Apgi4jE5Ag1BpGdD6zgvWxkhGo7xQj7uSC3sSeXiuKC9wyj30lUFCJyi//mH27bReRCEck3c1NV/YaqBrkPfgAc7X8+Bfh3VX1MVaeBO/G8rIxFxOgokfYIgKPLO9MZxEkVUJ4Jnax7YRjdIM3S01fxkgBO+O1LeMtRvwEuzWAMb/DvAXAUsCt07Ff+vjpEZJ2I7BCRHffdd18GwzB6hU2b4FfSWrWl9eth69bqOuFbt7ZHWWRa98IwepA0iuJlqnqeqt7it43Ai1X1/cCY6yIRuVpEfhrRTgmdsxE4ADT87qWqk6q6WlVXH3744Y1ebvQwExOw89xNPCLN+69OTja2vxXSBO8ZRj+TJuAuJyLPU9XrAUTkjyBIn4kzdaaqviyuUxFZC7wKWKMLrld3A6tCpx3t7zMWGSdtmYAX0ECUXjUpA8ozwYL3jEEnzYzibGCbiEyLyDSwDThHREaA/9PMTUXkFcD/Bl6jquFJ+5XA60XkIBE5BngacH0z9zAGgIai9KpxlUmNK5/aLCef3HzdC8PoBxJnFKr6I+C/isiT/O2HQoc/3eR9LwIOAq4S73/YD1T1XFW9VUQ+DfwMb7byRlVtwzugMeisW+fZJKL2Z8nUFFx2WXXtChE488ze8cgyjFZJE3D3FOC9wJGq+krfhfVEVd3WiQGmwQLujCjWr/dsEvPz3kxi3brsA8pdBZssENDoB9IG3KVRFF8FPgFsVNU/EJElwE9U9b9mM9TWMUVhdAtnJTzxVswMo5fJMjL7MFX9NF7xIvz4B1sOMgxiKuGZIdsYINIoin1+YJ0CiMjzgYfiLzGMxYFloTUWA2kUxVvxvJGeKiLfBS6HoPivYXSPXkibYVlojcVAoo0CwLdLPB0Q4A5VnWv3wBrBbBSLj9raRuC9yZuQNoz0tGyjEJE/EpH/AhW7xHOBTcAHRWRlZiM1jCawtBmG0Tnilp4+BjwOlbrZ78NbdnoIaEMiBMNIj6XNMIzOEacocqr6gP/5b4FJVf2cqp4PHNv+oRmGG/M2MozOEasofNsEwBrgm6FjaXJEGUbbMG8jw+gccYriU8C3ROSLwH5gO4CIHIu5xxpdxryNDKNzxHo9+TETRwDfUNV9/r7fB1ao6o87M8RkzOvJMAyjcdJ6PcUuIanqDyL2/byVgRmGYRj9RZqAO8MwDGMRY4rCMAzDiMUUhWEYhhGLKQrDMAwjFlMUhmEYRiymKAzDMIxYTFEYhmEYsZiiMAzDMGIxRWEYhtEL9EIlLgddURQi8h4RuVlEbhSRb4jIkf5+EZGPiMid/vETujE+wzCMjhJU4pqdBVXv77p1PaMsujWj+ICqPkdVjwe+DLzD3/9K4Gl+Wwds7dL4DMMwOkePV+LqiqJQ1YdDmyNAkJnwFOBy9fgBcIiIHNHxARqGYXSSHq/E1TUbhYhsEpFdwAQLM4qjgF2h037l74u6fp2I7BCRHffdd197B2sYhtFOerwSV9sUhYhcLSI/jWinAKjqRlVdBUwBb2q0f1WdVNXVqrr68MMPz3r4hmEYnaPHK3G1rVKdqr4s5alTwFeAdwJ3A6tCx4729xmGYQwuQcWtjRu95abRUU9J9Eglrm55PT0ttHkKcLv/+UrgDN/76fnAQ6p6T8cHaBiG0WkmJmBmBspl72+SkuigO223al+/T0SeDpSBWeBcf/9XgJOBO4FHgLO6MzzDMIweJnCnDTylAndaaMssJLYUar9gpVANw1hUjI15yqGWQsGbjaQkbSlUi8w2DMPoNzrsTmuKwjAMo9/osDutKQrDMIx+o8PutKYoDMMw+o2JCZic9GwSIt7fycm2udN2y+vJMAzDaIWJiY7FWdiMwjAMw4jFFIVhGIYRiykKwzAMIxZTFIZhGEYspigMwzCMWAYihYeI3IeXMyqKw4D7OzicTjGIzzWIzwT2XP3EID4TuJ+roKqJdRoGQlHEISI70uQy6TcG8bkG8ZnAnqufGMRngtafy5aeDMMwjFhMURiGYRixLAZFMdntAbSJQXyuQXwmsOfqJwbxmaDF5xp4G4VhGIbRGothRmEYhmG0gCkKwzAMI5aBVRQi8i4RuVtEbvTbyaFj54nInSJyh4j8eTfH2Sgi8gp/3HeKyNu6PZ5WEJEZEbnF/312+PtWishVIvIL/++h3R5nEiJyiYjsFpGfhvZFPod4fMT//W4WkRO6N3I3jmfq+/9TIrJKRK4VkZ+JyK0i8hZ/f9/+XjHPlN3vpaoD2YB3AX8fsf844CbgIOAY4JdArtvjTflMOX+8vwcs85/juG6Pq4XnmQEOq9l3AfA2//PbgPd3e5wpnuNFwAnAT5OeAzgZ+CogwPOBH3Z7/A08U9//nwKOAE7wPx8M/Nwff9/+XjHPlNnvNbAzihhOAf5dVR9T1WngTuB5XR5TWp4H3Kmqd6nq48C/4z3PIHEKcJn/+TLgL7s4llSo6reBB2p2u57jFOBy9fgBcIiIHNGZkabH8Uwu+ub/lKreo6o/9j/vAW4DjqKPf6+YZ3LR8O816IriTf508ZLQEsZRwK7Q4VRM4QAABrdJREFUOb8i/kvtJfp57FEo8A0RuUFE1vn7nqKq9/iffwM8pTtDaxnXc/T7bzgw/6dEZAz4Q+CHDMjvVfNMkNHv1deKQkSuFpGfRrRTgK3AU4HjgXuAD3Z1sEYUJ6nqCcArgTeKyIvCB9WbJ/e9//agPAcD9H9KRFYAnwP+h6o+HD7Wr79XxDNl9nv1dSlUVX1ZmvNE5OPAl/3Nu4FVocNH+/v6gX4eex2qerf/d7eIfB5v+nuviByhqvf4U/zdXR1k87ieo29/Q1W9N/jcz/+nRGQpnkCdUtX/9Hf39e8V9UxZ/l59PaOIo2Yd8a+AwHvjSuD1InKQiBwDPA24vtPja5IfAU8TkWNEZBnwerzn6TtEZEREDg4+Ay/H+42uBM70TzsT+GJ3Rtgyrue4EjjD96Z5PvBQaMmjpxmE/1MiIsA24DZV/VDoUN/+Xq5nyvT36rbFvo2eAFcAtwA3+1/MEaFjG/Es/XcAr+z2WBt8rpPxvBp+CWzs9nhaeI7fw/O8uAm4NXgWIA9cA/wCuBpY2e2xpniWT+FN7efw1nv/zvUceN4zH/V/v1uA1d0efwPP1Pf/p4CT8JaVbgZu9NvJ/fx7xTxTZr+XpfAwDMMwYhnYpSfDMAwjG0xRGIZhGLGYojAMwzBiMUVhGIZhxGKKwjAMw4jFFIXRMiIy72en/KmIfEZEhh3nfa/TYxsEROTtbez7NdJkFmIR+WcRSRX0GnHtWhG5qJlrjc5j7rFGy4jIXlVd4X+eAm7Q6sCfJap6oE33blvfvUL4+8243659dyKyFi8m4U3duL/RGDajMLJmO3CsiLxERLaLyJXAz8ATeP7fl4jIt0TkiyJyl4i8T0QmROR68epTPNU/73AR+ZyI/MhvL/D3v0tErhCR7+IFFVUhIv/o93OTiLzP33e8iPzAT5D2eVmoN3CdiFwoIjtE5DYR+SMR+U/x6hL8i3/OmIjcLiJT/jmfDWZNIrJGRH7i3+8SETnI3z8jIu8WkR/7x57h7x/xz7vev+4Uf/9a/75f8+99gb//fcByf8Y25e87zb/+RhH5mIjkIr6DGRG5wL/39SJyrL//UhG5WER+CFwQfrP3j31ERL7n/y6vS/hOLw3Oibnfq0Xkh/6zXi0i/ZrkcXHT7ahCa/3fgL3+3yV4qQ+KwEuAfcAxEee9BPgdXh79g/DyzLzbP/YW4MP+50/iJQ4EGMVLUQBenv0bgOURY3kl8D1g2N8OImxvBl7sf/7n0D2uY6H2wFuAX4fG9Su8iN0xvMjXF/jnXQL8PfAEvCycv+/vvxwvIRt4tTY2+J/XA//X//xe4DT/8yF4UfYjwFrgLuBJfr+zwKrw9+Z/fibwJWCpv70FOCPie5hhIdr9DODL/udL8XL+5PzttcBFoWOfwXuBPA4vpX3cd3op8LqE+x3KwsrF2cAHa+9rrfebzSiMLFguIjcCO4CdeHlnAK5XL999FD9SL4/+Y3ipBL7h778FTzADvAy4yO/7SuCJ4mXIBLhSVfdH9Psy4BOq+giAqj4gIk8CDlHVb/nnXIZXmCcgyJd1C3BraFx3sZA8bZeqftf/XMJLm/B0YFpVf+7oN0g4d0PomV4OvM1/puvwlMKof+waVX1IVR/Fm4UVIp5vDfBc4Ed+H2vw0qFE8anQ3xND+z+jqvOOa76gqmVV/RkLqbbrvtMG7nc08HURuQX4B+BZjmuNHqavs8caPcN+VT0+vENEwJtRuHgs9Lkc2i6z8O9yCHi+Lzgb6btRwveuHVcwllpjXhrjXtDXfKgfAV6rqneETxSRP665d/iaqlOBy1T1vBT3V8fntL+LpLhH0v02Ax9S1StF5CV4s0Gjz7AZhdHLfAPYEGyIyPEx5wZcBZwVsiGsVNWHgAdF5IX+OacD33J14GBURIK35FOB7+AlVBsL1uNT9vt1YIP42k5E/jDFvefESyMNXuK614nIk/3rV4pI1MwD4G9Df7+f4j4u6r7TBu73JBZSWJ9Zd4XRF9iMwuhl3gx8VERuxvu3+m3g3LgLVPVrvkLZISKPA18B3o4npC72hd1dwFkNjuUOvOJKl+AtC21V1UdF5CzgMyKyBC8N/MUJ/bwH+DBws4gMAdPAqxKumfTP/7GqTojIP+FVBhzCy+76RjybRi2H+t/dY8B4usesJ+Y7TXO/d+F9Pw8C38Sr0Wz0GeYeaxgJiFde8suq+uwuDyU1IjKD5356/yDez+gstvRkGIZhxGIzCsMwDCMWm1EYhmEYsZiiMAzDMGIxRWEYhmHEYorCMAzDiMUUhWEYhhHL/wfBmDIrKNtcTQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "45weTvdO4aBK",
        "outputId": "5d18644e-1218-4d2a-ce48-a8e3d88045d2"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(X_pca, y_binary, test_size=0.3, random_state=1, stratify=y_binary)\n",
        "print('Datos totales:', np.bincount(y_binary))\n",
        "print('Datos en y_train:', np.bincount(y_train))\n",
        "print('Datos en y_test:', np.bincount(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos totales: [855 744]\n",
            "Datos en y_train: [598 521]\n",
            "Datos en y_test: [257 223]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rAMgd5Pm4dIQ",
        "outputId": "d21df643-863a-49a9-ce12-5a8ba3b621df"
      },
      "source": [
        "sc = StandardScaler()\n",
        "sc.fit(X_train)\n",
        "X_train_std = sc.transform(X_train)\n",
        "X_test_std = sc.fit_transform(X_test)\n",
        "ppn_pca = Perceptron(max_iter=40, eta0=0.1, random_state=1)\n",
        "ppn_pca.fit(X_train_std, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Perceptron(alpha=0.0001, class_weight=None, early_stopping=False, eta0=0.1,\n",
              "           fit_intercept=True, max_iter=40, n_iter_no_change=5, n_jobs=None,\n",
              "           penalty=None, random_state=1, shuffle=True, tol=0.001,\n",
              "           validation_fraction=0.1, verbose=0, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TkAWE-jB4jRm",
        "outputId": "f9690dcb-b951-4b92-e8c3-580164c4e167"
      },
      "source": [
        "y_pred = ppn_pca.predict(X_test_std)\n",
        "print('Datos mal clasificados: %d' % (y_test != y_pred).sum())\n",
        "print('Precisión: %.2f' % ppn_pca.score(X_test_std, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos mal clasificados: 194\n",
            "Precisión: 0.60\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "id": "POzWhrZv6_zn",
        "outputId": "e9ef081c-aad8-406f-b46f-2f727605f7e6"
      },
      "source": [
        "plot_decision_regions(X=X_train_std, y=y_train,\n",
        "                      classifier=ppn_pca)\n",
        "plt.xlabel('Primer componente principal')\n",
        "plt.ylabel('Segundo componente pricipal')\n",
        "plt.legend(loc='upper left')\n",
        "\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L56gBGjoxYxk"
      },
      "source": [
        "### Datos espectográfia óptica de fluorescencia para detección de CCU\n",
        "Los datos corresponden a 152 mujeres patológicas y no patológicas, los vectores por individuo tienen una longitud de 36 características que corresponden al análisis de la señal de espectroscopia. Esta señal se compone de 5 formas gaussianas y cada forma presenta características que se asocian con la presencia de cáncer de cuello uterino.  Para el ejercicio se cuenta con datos etiquetados por un experto, en las categorías de patológico o no."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ajAZBRl2xkaC",
        "outputId": "18511c61-9290-40d6-9384-b736f18f3538"
      },
      "source": [
        "import scipy.io as sio\n",
        "data = sio.loadmat('/content/drive/MyDrive/Dataset/dataCancer.mat')\n",
        "data = data['X']\n",
        "x = data[:,0:-1]\n",
        "y = []\n",
        "for label in data[:,-1]:  y.append(int(label)) \n",
        "print('Numero de clases - Datos CCU:', np.unique(y))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Numero de clases - Datos CCU: [1 2]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gysOIEUazkpG"
      },
      "source": [
        "Para estos datos se hará primero la extracción de las componentes principales, pero es necesario previamente estandarizar los datos. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k9E1WG8-zkRm"
      },
      "source": [
        "sc = StandardScaler()\n",
        "sc.fit(x)\n",
        "x_std = sc.transform(x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "niwjv-tx0Ijt"
      },
      "source": [
        "pca = decomposition.PCA(n_components=2)\n",
        "pca.fit(x_std)\n",
        "X_pca = pca.transform(x_std)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "as0iwgjg0Tvp",
        "outputId": "4f138278-48f0-42d2-ba52-316d7bc207fd"
      },
      "source": [
        "k = 0\n",
        "colors = ('red', 'blue')\n",
        "for point in X_pca:\n",
        "  plt.scatter(point[0],point[1], c=colors[y[k]-1])\n",
        "  k +=1\n",
        "#plt.scatter(X_pca[:,0],X_pca[:,1])\n",
        "plt.xlabel('Primer componente principal')\n",
        "plt.ylabel('Segundo componente principal')\n",
        "plt.title('PCA 2 componentes')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IYyF3fux14mW"
      },
      "source": [
        "Ahora se establecen la division de datos entre entrenamiento y prueba. \n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "01b6hDSJ2Ccv",
        "outputId": "040a5407-fcea-4d33-e566-9055627a2938"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(X_pca, y, test_size=0.3, random_state=1, stratify=y)\n",
        "print('Datos totales:', np.bincount(y))\n",
        "print('Datos en y_train:', np.bincount(y_train))\n",
        "print('Datos en y_test:', np.bincount(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos totales: [ 0 76 76]\n",
            "Datos en y_train: [ 0 53 53]\n",
            "Datos en y_test: [ 0 23 23]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TpTLNq3r2dsU",
        "outputId": "424f6ba2-3d9a-4be6-bdae-db19f94fab2a"
      },
      "source": [
        "ppn_ccu = Perceptron(max_iter=10, eta0=0.2, random_state=1)\n",
        "ppn_ccu.fit(X_train, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Perceptron(alpha=0.0001, class_weight=None, early_stopping=False, eta0=0.2,\n",
              "           fit_intercept=True, max_iter=10, n_iter_no_change=5, n_jobs=None,\n",
              "           penalty=None, random_state=1, shuffle=True, tol=0.001,\n",
              "           validation_fraction=0.1, verbose=0, warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NxyN0lnC5Qdu",
        "outputId": "b49821be-45a9-47cf-a520-d72c6f9b6caf"
      },
      "source": [
        "print('W1: %.3f;' % ppn_ccu.coef_[0][0], ' W2: %.3f' % ppn_ccu.coef_[0][1])\n",
        "print('Intercepto: %.3f' % ppn_ccu.intercept_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "W1: -1.160;  W2: -0.111\n",
            "Intercepto: -0.200\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "44-P_s3C2oFI",
        "outputId": "3acc32b6-5da1-4a6a-820d-e284d11c5e81"
      },
      "source": [
        "y_pred = ppn_ccu.predict(X_test)\n",
        "print('Datos mal clasificados: %d' % (y_test != y_pred).sum())\n",
        "print('Precisión: %.2f' % ppn_ccu.score(X_test, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos mal clasificados: 0\n",
            "Precisión: 1.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "id": "0D_nWVjA21Sc",
        "outputId": "227ff444-b949-44dc-8459-a5ccb70aacb1"
      },
      "source": [
        "plot_decision_regions(X=X_train, y=y_train,\n",
        "                      classifier=ppn_ccu)\n",
        "plt.xlabel('Primer componente principal')\n",
        "plt.ylabel('Segundo componente pricipal')\n",
        "plt.legend(loc='upper left')\n",
        "\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BzVxUMpik7b3"
      },
      "source": [
        "## Clasificadores No lineales"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eFTr2Y04xzup"
      },
      "source": [
        "En esta sección se realizará la construcción de un clasificador no lineal, en este caso se probará con un perceptron multicapa de la libreria Sklearn. El conjuntos de datos para esta fase corresponde a una aplicación de clasificación de vidrios, se cuenta con X datos divididos en 2 clases: vidrios para la contrucción de ventanas procesados y no procesados. El vector de caracteristicas se compone de 9 atributos numericos :\n",
        "\n",
        "* RI: Indice de refracción\n",
        "* Na: Sodio presente en el vidrio, unidad de medida porcentaje de peso presente en oxido\n",
        "* Mg: Magnesio\n",
        "* Al: Aluminio\n",
        "* Si: Silicon\n",
        "* K: Potasio\n",
        "* Ca: Calcio\n",
        "* Ba: Bario\n",
        "* Fe: Hierro\n",
        "\n",
        "Base de datos: https://archive.ics.uci.edu/ml/datasets/glass+identification"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "udX4l6jCk7D5",
        "outputId": "1c6c7e91-de65-4632-b8b8-519604b6badb"
      },
      "source": [
        "import scipy.io as sio\n",
        "data = sio.loadmat('/content/drive/MyDrive/Dataset/data_glass.mat')\n",
        "data = data['data']\n",
        "x = data[0:146,0:-1]\n",
        "y = []\n",
        "for label in data[0:146,-1]:  y.append(int(label)) \n",
        "print('Numero de clases - Datos Clasificación de vidrio:', np.unique(y))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Numero de clases - Datos Clasificación de vidrio: [1 2]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uOezaGiCzath"
      },
      "source": [
        "El anterior bloque de codigo importa los datos y presenta el número de etiquetas presentes."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "x-Lnz-f5lr4E",
        "outputId": "26bcae90-88ba-42dc-e9b9-6f1b4073efcf"
      },
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.3, random_state=1, stratify=y)\n",
        "print('Datos en y_train:', np.bincount(y_binary))\n",
        "print('Datos en y_train:', np.bincount(y_train))\n",
        "print('Datos en y_test:', np.bincount(y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos en y_train: [855 744]\n",
            "Datos en y_train: [ 0 49 53]\n",
            "Datos en y_test: [ 0 21 23]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vX6p8hvflsk7"
      },
      "source": [
        "sc = StandardScaler()\n",
        "sc.fit(X_train)\n",
        "X_train = sc.transform(X_train)\n",
        "X_test = sc.transform(X_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hn7P6GGXzsJW"
      },
      "source": [
        "Se construye un clasificador perceptron multicapa con una capa oculta y de 5 neuronas, se indica un maximo de iteraciones en 500."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9xBM4qRaojB2",
        "outputId": "27be31a8-6933-4203-9ddf-4f14b4b54ae8"
      },
      "source": [
        "from sklearn.neural_network import MLPClassifier\n",
        "clf = MLPClassifier(solver='lbfgs', alpha=1e-5, hidden_layer_sizes=(5,), random_state=1, max_iter=500)\n",
        "clf.fit(X_train, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "MLPClassifier(activation='relu', alpha=1e-05, batch_size='auto', beta_1=0.9,\n",
              "              beta_2=0.999, early_stopping=False, epsilon=1e-08,\n",
              "              hidden_layer_sizes=(5,), learning_rate='constant',\n",
              "              learning_rate_init=0.001, max_fun=15000, max_iter=500,\n",
              "              momentum=0.9, n_iter_no_change=10, nesterovs_momentum=True,\n",
              "              power_t=0.5, random_state=1, shuffle=True, solver='lbfgs',\n",
              "              tol=0.0001, validation_fraction=0.1, verbose=False,\n",
              "              warm_start=False)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gi10s2Cmot_p",
        "outputId": "d1ac61ca-3ed2-4643-b876-cda20985ecd7"
      },
      "source": [
        "y_pred = clf.predict(X_test)\n",
        "print('Datos mal clasificados: %d' % (y_test != y_pred).sum())\n",
        "print('Precisión: %.2f' % clf.score(X_test, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Datos mal clasificados: 11\n",
            "Precisión: 0.75\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3etPTYU3z55M"
      },
      "source": [
        "La primera validación se realiza al variar el número de capas ocultas del perceptron, se toman capas desde 1 hasta 7 y en todas se les ubican 9 unidades neuronales."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "KUXXIL_ao0Zr",
        "outputId": "a8ebe7a1-9110-470a-ca37-80c39d3f909b"
      },
      "source": [
        "from sklearn.model_selection import validation_curve\n",
        "param_range = [(9,)*1, (9,)*2, (9,)*3, (9,)*4, (9,)*5, (9,)*6, (9,)*7]\n",
        "param_range_1d = [1, 2, 3, 4, 5, 6, 7]\n",
        "train_scores, test_scores = validation_curve(\n",
        "                estimator=clf, \n",
        "                X=X_train, \n",
        "                y=y_train, \n",
        "                param_name='hidden_layer_sizes', \n",
        "                param_range=param_range,\n",
        "                cv=10)\n",
        "\n",
        "train_mean = np.mean(train_scores, axis=1)\n",
        "train_std = np.std(train_scores, axis=1)\n",
        "test_mean = np.mean(test_scores, axis=1)\n",
        "test_std = np.std(test_scores, axis=1)\n",
        "\n",
        "plt.plot(param_range_1d , train_mean, \n",
        "         color='blue', marker='o', \n",
        "         markersize=5, label='Precisión entrenamiento')\n",
        "\n",
        "plt.fill_between(param_range_1d , train_mean + train_std,\n",
        "                 train_mean - train_std, alpha=0.15,\n",
        "                 color='blue')\n",
        "\n",
        "plt.plot(param_range_1d , test_mean, \n",
        "         color='green', linestyle='--', \n",
        "         marker='s', markersize=5, \n",
        "         label='Precisión test')\n",
        "\n",
        "plt.fill_between(param_range_1d , \n",
        "                 test_mean + test_std,\n",
        "                 test_mean - test_std, \n",
        "                 alpha=0.15, color='green')\n",
        "\n",
        "plt.grid()\n",
        "plt.legend(loc='lower right')\n",
        "plt.xlabel('Número de capas ocultas')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.ylim([0.3, 1.0])\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:470: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
            "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
            "\n",
            "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
            "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
            "  self.n_iter_ = _check_optimize_result(\"lbfgs\", opt_res, self.max_iter)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b-HqaHkm0QQN"
      },
      "source": [
        "Ahora se realizará el mismo procedimiento pero variando el número de neuronas en la capa oculta, se tomara fijo 2 capas y se varia entre 2 y 40 neuronas en cada capa."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "id": "KVJLdcpQqJ-D",
        "outputId": "fdd78228-a9aa-4c54-985e-82602d456f9e"
      },
      "source": [
        "param_range = [(2,)*2, (5,)*2, (10,)*2, (15,)*2, (20,)*2, (25,)*2, (30,)*2, (40,)*2]\n",
        "param_range_1d = [2, 5, 10, 15, 20, 25, 30, 40]\n",
        "train_scores, test_scores = validation_curve(\n",
        "                estimator=clf, \n",
        "                X=X_train, \n",
        "                y=y_train, \n",
        "                param_name='hidden_layer_sizes', \n",
        "                param_range=param_range,\n",
        "                cv=5)\n",
        "\n",
        "train_mean = np.mean(train_scores, axis=1)\n",
        "train_std = np.std(train_scores, axis=1)\n",
        "test_mean = np.mean(test_scores, axis=1)\n",
        "test_std = np.std(test_scores, axis=1)\n",
        "\n",
        "plt.plot(param_range_1d , train_mean, \n",
        "         color='blue', marker='o', \n",
        "         markersize=5, label='Precisión entrenamiento')\n",
        "\n",
        "plt.fill_between(param_range_1d , train_mean + train_std,\n",
        "                 train_mean - train_std, alpha=0.15,\n",
        "                 color='blue')\n",
        "\n",
        "plt.plot(param_range_1d , test_mean, \n",
        "         color='green', linestyle='--', \n",
        "         marker='s', markersize=5, \n",
        "         label='Precisión test')\n",
        "\n",
        "plt.fill_between(param_range_1d , \n",
        "                 test_mean + test_std,\n",
        "                 test_mean - test_std, \n",
        "                 alpha=0.15, color='green')\n",
        "\n",
        "plt.grid()\n",
        "plt.legend(loc='lower right')\n",
        "plt.xlabel('Numero de neuronas configuración 2 capas ocultas')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.ylim([0.5, 1.1])\n",
        "plt.tight_layout()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3O7T53gv1FHB"
      },
      "source": [
        "Para probar el comportamiento del clasificador cuando se varia el número de neuronas en las capas ocultas y el número de capas, se realiza el siguiente ciclo en el cual se estan variando ambos hiperparametros."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 993
        },
        "id": "LXW4aLh3sXZ9",
        "outputId": "b4d0fd88-55cb-42d9-ef33-b6009b9f5621"
      },
      "source": [
        "\n",
        "neuron_vector = [2, 6, 10, 15, 20, 25]\n",
        "k = 0 \n",
        "plt.figure(figsize=(25,4)) \n",
        "for neuron_unit  in neuron_vector:\n",
        "  if k == 0: \n",
        "    fig, axs = plt.subplots(1, 2) \n",
        "    \n",
        "  param_range = [(neuron_unit,)*1, (neuron_unit,)*2, (neuron_unit,)*3, (neuron_unit,)*4, (neuron_unit,)*5, (neuron_unit,)*6, (neuron_unit,)*7, (neuron_unit,)*8]\n",
        "  param_range_1d = [1, 2, 3, 4, 5, 6, 7, 8]\n",
        "  train_scores, test_scores = validation_curve(\n",
        "                  estimator=clf, \n",
        "                  X=X_train, \n",
        "                  y=y_train, \n",
        "                  param_name='hidden_layer_sizes', \n",
        "                  param_range=param_range,\n",
        "                  cv=15)\n",
        "\n",
        "  train_mean = np.mean(train_scores, axis=1)\n",
        "  train_std = np.std(train_scores, axis=1)\n",
        "  test_mean = np.mean(test_scores, axis=1)\n",
        "  test_std = np.std(test_scores, axis=1)\n",
        "\n",
        "  axs[k].plot(param_range_1d , train_mean, \n",
        "          color='blue', marker='o', \n",
        "          markersize=5, label='Precisión entrenamiento')\n",
        "\n",
        "  axs[k].fill_between(param_range_1d , train_mean + train_std,\n",
        "                  train_mean - train_std, alpha=0.15,\n",
        "                  color='blue')\n",
        "\n",
        "  axs[k].plot(param_range_1d , test_mean, \n",
        "          color='green', linestyle='--', \n",
        "          marker='s', markersize=5, \n",
        "          label='Precisión conjunto de pruebas')\n",
        "\n",
        "  axs[k].fill_between(param_range_1d , \n",
        "                  test_mean + test_std,\n",
        "                  test_mean - test_std, \n",
        "                  alpha=0.15, color='green')\n",
        "  \n",
        "  axs[k].grid()\n",
        "  axs[k].legend(loc='lower right')\n",
        "  axs[k].set_title('Presición vs número de capas para %d neuronas' % neuron_unit)\n",
        "  axs[k].set_xlabel('Número de capas ocultas')\n",
        "  axs[k].set_ylabel('Precision')\n",
        "  axs[k].set_ylim([0.4, 1.1])\n",
        "  k += 1\n",
        "  if k>= 2: k= 0\n",
        "  #axs[k].set_tight_layout()\n",
        "  #axs[k].show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:470: ConvergenceWarning: lbfgs failed to converge (status=2):\n",
            "ABNORMAL_TERMINATION_IN_LNSRCH.\n",
            "\n",
            "Increase the number of iterations (max_iter) or scale the data as shown in:\n",
            "    https://scikit-learn.org/stable/modules/preprocessing.html\n",
            "  self.n_iter_ = _check_optimize_result(\"lbfgs\", opt_res, self.max_iter)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1800x288 with 0 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAEXCAYAAADsu06BAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9eZgcVdX4/zm9TS+zJpMESEI2EiBsgQQIYQsQVkERRXB9VRb1FVxQkRcXcP3qC+KGovBGI7IJiv6QVbYkkLCHQCAhIclkn33vfbu/P271TE9P96w9PT2T+3mefma6qrrqVNWpOveee865opTCYDAYDAZD4bCNtgAGg8FgMOxvGONrMBgMBkOBMcbXYDAYDIYCY4yvwWAwGAwFxhhfg8FgMBgKjDG+BoPBYDAUmKIwviJyo4j83wC2+4OIfC/L8ktF5CkRKRkZCfOLiFwnIvePthwDRURWiMiPR1uOsYjR7eLG6LZhtHD0t4GI7ACmAAkgADwBXKOU8udLCKXUTwe43RezyHcscCXwYaVUJF8yjRQiMhP4NHDa6EpiGE3dFpF5wC3AEsAOvAb8MmMbo9uGMYmI2IEfAJ8HyoCtwBlKqbZRFayIGGjP9yKlVClwHLAI+G7mBiLSryEfCZRSbyqlzlVKBUfj+EPgUODTSqnOQhxMNEXh4ShS8qLbQ9D/SuARtD5MAV4F/r/0DUZDt4f5HBvdHocMUSd+gG5YngSUoxtl4XzKNVhGy0blRCnV5wfYASxL+34L8Kj1vwK+DLwP1FjLLgTWA23AWuDotN9+G9gLdAKbgbOs5TcD96Rtd4r12zZgN/BZa/kK4Mdp212FblG1oF9kB6WtU8AXLdnagN8BkuX8DgJCwIS0ZccCTYATOARYBbRby/6W4zrNtI75X8Aua9vvpK3PlH0psCfjOn8LeBvdC1uOfik/YV2vZ4CqtO0Xp12jt4ClaetWAj8B1ljndgj6QXjNOo/XgCV93PNjgXXWcf8GPJAhe857nGVfRwBPW/eoHrjRWn4C8JK1j1rgdsCVcf++Amy3ruUtgM1aNwd4Dmi21t0LVPanZ1lk8wP/tuTrROvas2nHfx6IAUngDeD6tPPebf3uHqAD+DvQgO5FJyzZbgd+xMB0+z7rmBNz6Pbf0Dr8mLVdjaUzKd0+LO06bwY+RrduvwhcmXZvOy3dSOm2sq5FJ93Pcfrxn6anbit0QyH1bG1K6Yd1b9ZZ1yB1b3bRrdtRtH77rU8Qo9sjodsrgD/QrdurgBlp63+N1r8OtG6fmrbuZrQ+p3T7yv7OKePYVda9ndOffUmTNaXbncAr6b8li25n6MOVad8/C7yYca0zbdSQ7Ea+7k3X9gO4MDuwjC8wHXgX+FGaoE8DEwAPWrEbgBPRrrT/sn5fgm4V706dKNpYzUm72fdY/8+whP842vhNBBZkGjDgTOsCHGft/7fA6oyL+Ci6h3Ew0Aicl+McnwOuSvt+C/AH6//7ge+gvQRu4JQc+5hpHfMu61ocA0SAwzNlt74vpbfxfRltcKda13GddU3dlow3WdtOtRTgAkuus63vk9IUchf65eCw9tmKbn06rGvbivWizzgPF7AT+Lp1/T+KNkCp657zHmfZVxn6Qf2GdQ5lwInWuoXol6zDunabgK9l3L/n0bp1MLCFbgNyiHXOJcAkYDXwK2tdTj3LIl/q5X8a+sFqBnalHX+DtbwU+AXaoJxqnfc/rW0+in457AHOtc5pjrXPTcCTDEy3nwM6+9DtOku+E6zj7gH+QbduNwKfs65nqvE439rv5rRrd4t1XV+kW7cV8CxwPlp3M4+/gp66raxrcYJ1/DBwX9q9+YYlX+redKB1ewmwz9r3OrT+Ho7R7ZHQ7RVoXTvN2tev6WmUPoXWP4d1DnWAO+19HAMutu6Bp79zyjj2aWjD9W1rv1uAL/djfFO67UAbtQesdT7rnHvp9iCMb7qNGrLdyNe9Gazx9VsXcyfwe8CTJuiZadvegWWY05ZtBk63BG8AlgHOjG1upvsF9T/AP/u4SakHZTnwv2nrSi2FmZkm2ylp6x8Ebsix3yuB56z/xbqIp1nf7wbuBKb1c51mWseclrbsVeDyTNmt70vpbXw/mfb9H8Adad+vBf5l/f9t4K8Zx38K+K80hfxh2rpPA69mbP8SVq8ry4OzjzQvAboHkLruOe9xln19HHizPx2ztv1a+n23ruV5ad//G6tXmuW3F6eO05eeZfmd39KZlG7fiTYq08mu2yHgmDSdDdKHblvntIl+dBuYhu4N/rMP3U7S/UJSwDeB99LuZU3GPv8I3ITW7Vbrb0q3f4o2vindzjzXzOPPt7ZZnHb8jXTrdg3W85Op29a9iQCfTLtOLwB/NLo9orq9IqUvaTqUAKbn2L6Vnrq9ejDnlLHuE9Y5LkcbvKPRRuzsPmT9v7TvF6Tp9mXAC9l0O00f+jO+fen2cOzGkO5N6jPQ8ZKLlVKVSqkZSqn/VkqF0tbtTvt/BvANEWlLfdAvsoOUUlutG3Yz0CAiD4jIQVmONR3YNgCZDkK/MAFQOkimGd1yTlGX9n8QfaGz8Q/gJBE5EP2AJtEvCNCuRgFeFZF3ReTz/cg10GNmoz7t/1CW76l9zQAuzbjOpwAHpm2ffl96XCuLnfS8Vunb7lWWRqVtmyLnPc6yr5z3UkTmicijIlInIh1og1CdsVn6OexMHUNEplj6s9f67T2p3w5Cz1L8M023r0a7olLbnyMim0SkHW283MAa65xvQPek0nX7Z0BAREIi0mmdk7ev6yEik4D/oF/y71qLs+l2hJ5jZrV064MTODjjnnwSOACt2+VABd26vcX6XUq3Af6YptuZ+pIac3alLetMO37CkgERmQJ8Hzgw7d7Ygfq063Qo8Jm0e2N0e2R0u+sYlg516baIfDOl29Z5VmTImC7fQM8pRco+/FApFVJKvY1271/Qh6y53pszgBNz6PZAyakvg7Ebeb43eUk1Slfk3cBPrJdZ6uNVSt1vCXifUuoU9AVVwM+z7G832m3XH/us/QAgIj60G2XvoE9AqVb0C/AydKvtgdQDqpSqU0pdpZQ6CPgC8HsROWSwx0D3bNJfxINRnkx2o3sH6dfZp5T6Wdo26felx7WyOJjs16oWmCoikrFt+rFz3uMscs7OcQ53AO8Bc5VS5cCNdBuCFNMzZNhn/f9T9PkdZf32U+m/HaCepZiU+kdEStHuqdRxrkKPnVYBf0Ibv4uVUpVoQ/tAum6jxxz/gHZb/SvLOfXQbRGpQuvdI+jx0BTZdLsEbfCyEQB2ZtyTUqXUlyzdbkS7fD+BfgkeYMlcp5S6ytrHd+nW7Ux98Vh/cz1bMSzji743JUBd2r3pwrpO/wF+RfZ7Y3Q7f7rddYx03RaRU9ENr4+hx9or0bqbLmP6NR7oOaVI6XL6PjL3N1B2A6uy6ba1fiDv1Zz6Mki7kc97k/c837uAL4rIiVYkok9EPiAiZSJyqIicaeUrhtGto2SWfdwLLBORj4mIQ0QmisiCLNvdD3xORBZY+/wp8IpSascQZb8P+Ax6HOi+1ELReZbTrK+t6IuaTe7+WA9cICITROQAdCtpqNwDXCQi54qIXUTcIrI0Tc5MHgfmicgnrGt6GdqV+GiWbV8C4sBXRMQpIpegx2JS5LzHWfb1KLoH9DURKbH04ERrXRl6LNAvIocBX8ry+2+JSJWITAe+ig6QSf3WD7SLyFR0MA8Ag9CzFCeIyCki4kIHR72slEq1lONow+VA33c3cJj18nYCB6XrNrqH2Up3rzDznNJ1uwo9zrpZKXVDxnbZdLsR/YLMxm5ggoh82rpnThE5XkQOt9avBc5B6/YLwBXWtUrX7Xa6dTvz+Knrm9nDTNECHCIiE9CNmUMAleveoN87cbLfG6Pb+dPtC3Lodhlpui0i30frbl8M5JwAUEptQ+vZd6xrczhwOdnvSX88ir6/uXR7PXCJiHithuMV/exvOHYjn/cmv8ZXKfU6urdwO/oltBXtgwfdGv4ZerC7DpiMHgPL3McutHviG+iHej06eClzu2eA76HdarXoHsXlwxD/EWAuusX+Vtry44FXRMRvbfNVpdT2Iez/r+jIzR3olv/f+ty6D6wH6EPo1mcj+uX7LXLcT6VUMzqK8xtoF8v1wIVKqaYs20aBS9D3rQXtDXg4bX1f9zhzX53oAIWL0Pf8feAMa/U30T2xTvRLL9v1+P/QkZjr0ZGQy63lP0AHTLRbyx9O+82A9CyN59Bjoy3ooJL0ntpqtIt2JzoytQE9PtmKjlZN9WJTxzwC3YP8DNq91eOcMnR7Hzp45MOWbn0SuF5EDs6h26v6OIc4usF4ubXfOnSrO1WY4yvo3muldS3utZYfj44sBT2O9lWl1PYsx8/sWWayHT1EsgPtUo5a55/r3lyKvv+97o3R7bzq9n1k1+2n0IGAKd0Ok+FmzsJAzimdj6P1ptk6j+8ppZ7t5ze9sK7zOeTW7V+i9a0e+Avdup1rf8OxG/m8N10h1AZDUSEiCu3i2jqCx1iBDgzqldtrMIwURrcNUCTlJQ0Gg8Fg2J8YN8ZXRP4kIg0i8k6O9YeJyEsiEhGRbxZaPoNhqBjdNhjGH+PG7Swip6EHw+9WSh2ZZf1k9BjExUCrUurWAotoMAwJo9sGw/hj3PR8lVKr0YEFudY3KKVeQ6dFGAxjBqPbBsP4Y9wYX4PBYDAYxgrFNctDkSAiVwNXA3g8noXTp0/v5xf5IZlMYrMVb3uomOUrZtkgt3xbtmxpUkpNyvKTEcHodm+KWTYobvn6kq3Quj3WMMY3C0qpO9E1b1m0aJF6/fXXC3LclStXsnTp0oIcaygUs3zFLBvklk9EchWuGBGMbvemmGWD4pavL9kKrdtjjeJsThkMBoPBMI4ZNz1fEbkfPZtKtYjsQVd2cQIopf4guqTj6+gyakkR+Rp6WqqOURLZYBgQRrcNhvHHuDG+SqmP97O+Dj11m8EwpjC6bTCMP4zb2WAwGAyGAmOMr8FgMBgMBcYYX4PBYDAYCowxvgaDwWAwFBhjfA0Gg8FgKDDG+BoMBoPBUGCM8TUYDAaDocAY42swGAwGQ4ExxtdgMBgMhgJjjK/BYDAYDAXGGF+DwWAwGAqMMb4Gg8FgMBQYY3wNBoPBYCgwxvgaDAaDwVBgjPE1GAwGg6HAGONrMBgMBkOBMcbXYDAYDIYCY4yvwWAwGAwFxhhfg8FgMBgKjDG+BoPBYDAUGGN8DQbDfo1Soy2BYX/EGF+DwbDf0tkJwSDU1EAyOdrSGPYnxpXxFZE/iUiDiLyTY72IyG9EZKuIvC0ixxVaRoNhsIxlvVYKYrHRliI7wSBs2gQ2GzQ0wPvvF6+shvGHY7QFyDMrgNuBu3OsPx+Ya31OBO6w/hoMxcwKikivlYJEQn/icf1JJCAahUik59+UMQuH9XaOInnjRCLw3ntQUqKNb1UVdHTAxo1w6KHgdo+2hIbxTpE8CvlBKbVaRGb2scmHgLuVUgp4WUQqReRApVRtQQQ0GIZAofVaKfD7tbFMGdKUMU03qOnbgzZidnv3x+OB0lK9rrkZNm/Whm20DXAspmUBLWOK8nLdG96wAQ47DMrKRkc+w/7BuDK+A2AqsDvt+x5rmTG+hrFMXvU6FtM9QOhtUN1u8PkGv0+7XRu2LVtg3rzRM8CJBGzdqhsR5eW913u9et0772g5J04svIyG/YP9zfgOCBG5GrgaYMqUKaxcubIgx/X7/QU71lAoZvmKWTYoHvkGottKQSikDWa+iET8tLSspLERdu0aPbduJKINsN2ue+Mp2WpqVvbYTil4+WVwucDpLLyc6RSL7mSjmGUrdvY347sXmJ72fZq1rAdKqTuBOwEWLVqkli5dWhDhVq5cSaGONRSKWb5ilg1GXL4B6TUMTLejUVi/Hior8ydgTc1KZs3Sx2pv1+7ouXML1wNWShv92lqYMCG3bJm/aWmBSZNg1qz8NkYGQzHrdjHLVuyMq2jnAfAI8BkrOnQx0G7Gew3jgLzpdSIBjz0Gy5fD88/r7/mmokKPKb///sjsPxu1tbBvnw6sGigi2u3c0qKDs6LRkZPPsP8xrnq+InI/sBSoFpE9wE2AE0Ap9QfgceACYCsQBD43OpIaDAOnUHqdSMC558Irr0AgoIORjjlGG+J89/oqKqCtrXsMeCR7lY2NsHOnNrwig/99ZaVuLLzzjg7E8nrzL6Nh/2NcGV+l1Mf7Wa+ALxdIHIMhLxRKr594Qhtev19/Dwbhrbdg9Wo444zh7r03lZUjb4Db2nSAVWWlDh4bKqWlOl1qwwYdsZ1Pl7xh/2R/czsbDIYcvPmm7vGmEwrpQhQjRWWlrjK1dWv+XdB+vzbs5eX5MeypSO9Nm6C+fvj7M+zfGONrMBgAOPbY3mlEHg8cfvjIHjfVA86nAU41Gjye/EYru1xa3u3bYccOU5LSMHSM8TUYDACcfz6ceGJPA3z44XDaaSN/7KoqHQWdDwMcjeoAKadTV7DKN3a7jpiur9dBY/F4/o9hGP8Y42swGABtVJ56Cv76V7j8cp0GdMABhUuxqawcvgGOx7WrWamRDYwS0Q2Gzk549109HmwwDAZjfA0GQxd2O3zgA/D1r8MXv6jTjtauLdzxh2OAk0nYtk27nFNlLUea8nJ93A0btCE2GAaKMb4GgyErV18NM2bAD36gK0MViqGMASulx2Db2nQaUyHx+fTY8rvvdlfNMhj6Y1ylGhkMhuGjlCKcCFFZ4uX734crroC77oJrrimcDFVV0Nqqe7KHHNJ/mtCePXpawMzqVYXC5dK94M2b4aCDek7YINIzvzjze2pZru8mqGt8YoyvwWDoQSwZY3fgfaoqD+WUU9xccAH88Y9w0UW6J1woUgZ469a+DXBdHezePfqTIDgc2vg3NXXP9AQ9/8/1va/iH6GQblhMmjS0IiGG4sQYX4PB0AOXC2K2Tna272Ru1TxuuEFYtQp++EP4v/8rrAHozwC3tEBNjTZ6xWCYbLbssyUNh5QHoLMTZs4cvRrThvxixnwNBkMvqqtc1Le30BZpYcoU+NrX4MUXdRWsQpNugNNdsB0d2s1bUTG86lXFTqrGdHOzLnEZCo22RIZ8MI5V1mAwDJXSUvDay9kTqCGWjPGJT8ARR8BPf9pdfrKQpI8BJ5O69OV77+kJ70drbuBCU1mpXdQbNugev2FsY4yvwWDohbsEnHYnSQW1gd04HHDzzXo889e/Hh2Z0nvAmzbpco8u1+jIMlp4vTq6evNmPVmECcYauxjjazAYemGzaWNnj5XRFK7DH+vg6KN18Y177tFpNaNBKg3JbtfGdzgkEnraxHvvnTFi0yeOBE6nHuOuq9O9/0KmgRnyhzG+BoMhKxUVkEgIPkc5uzq3kUgmuO46bZRvvnn0jFVl5fCrVyUSOoXquuvgnntmct11+vtYMcCpClupmZY6OkZbIsNgMcbXYDBkJZWr6rK7iCWjNIT2UV4ON9wAb78NDz44uvINh9Wr9XSJwSAoJT2mTxxLlJZqD8C778K+fb3TmAzFizG+BoMhK06nfrlHIlDqrKAuuJdgPMBFF8HixXDbbXoMeCyycWPvqOFgUI9n33efXj9WJkxwuXQveOdOPdFDLDbaEhkGgjG+BoMhJ1VV2vjaxIbb4WGPvwZFku9/Xxuvn/98tCUcGvPn955q0G7XlbJ+8AP48Idh0SL41Kfgllvg6ad1oYtCMpgxaZtNpyN1dOh0pMx5mQ3Fx34SpF/8JFWSWCKGUgophmoBBgM6sjblynTbPXREW2kJNzJnzhSuuAL+8Af4yEd0T3gsceKJ+q9+1BQej3DMMbqISF2ddkGvX6///uUvejno0pHHHAMLFui/8+d3T1uYSGi39caNevlppw29IEZqTPqttyAUmsk//qGPt3x53/ssL+8eB54zR1fFMhQnxvgWCYFogGgiij/qp6ykbLTFMRgAbVhKSrQL1uEAn7OcPYEdlLkq+NKX3Dz2mA6+euSRsZX2c//9et7fr30NWlp2sGTJrC5jOW2a/nzgA3rbSESnNqWM8VtvdRcbcTr1nMdHHQWvvqp7zuGwHoedP1/3mpNJfaxYTH/i8e7/0z/RaPe6jRvh9ddTLuSeY9JnnNH3ubndWq6tW3VVrBkzTFWsYsQY3yKhNdSKQtEYaDTG11BUVFVBY6Me/7WLHYc42evfyazyeXzve8LVV+se2Ze+NNqSDozWVt1jP/10LXNNzU5mzZqVc/uSEt3TXbCge1lDgw46Sxnkhx7SxjNFKARvvAFnnpk/uUMh3Qjoz/iCNrapOtN+P8ybN/zULEN+Mca3CFBK0RRswmFz0Bhs5ODKg3HYzK0xFAdlZdoVm8LnLKUt0kx7tJXTT5/AuefCHXfAhRfC9OmjJ+dAueMObZC+9a2h72PyZFi2TH8Afvtb+N3vekcbn3UWnHOO7ommfxyO7v9drt7LX3kFvvvdnkFhHo/uZQ8UEZ2WFQjohsLcubohZSgOzBu+CAjFQ0QTUUQEpRQd4Q4meEdpbjSDIQOPRwf0JBXYrHCEUmc5u/3b8TnLuPFGJy+8AD/6kZ79qJhDFnbu1NHMl16qjVG+OPJIfZ2Cwe5lXq8+zkB6qpmcfz78/e/drme3W4/5nnba4Pfl8+l9bNqkG0dTp47vWthjhXF1C0TkPBHZLCJbReSGLOtniMizIvK2iKwUkWmjIWcm7eF2bNbT4HF6qA/Uj7JEhmJjNHXbZtMFNyLh7mUOmw4Vrg3s5oAD4KtfhVWrdFRwMfOLX+ie5bXX5ne/p52mjaPXqxsfXu/QjSVot/Hy5XDTTfr7hRf2H2zVF06njobetw+2bMlfOlI8GSeRHCOVSYqMcdPzFRE78DvgbGAP8JqIPKKU2pi22a3A3Uqpv4jImcD/Az5deGl70hBswOv00korHqeH1lArkXiEEkfJaItmKAKKQberqnRZx/RJ4n2OMprD9UxwV/OpT5Xzz3/Cj38MS5bo8eFiY906eOop+NKXo7gqAtQG/PhjnUTjQd5peQOnzYnD5sIpTpw2Fy5bCQ67E7vYsYk942/PfkvKWK5erXuYhx8+vGjn1D4vvRSWL/ezc2fpsIOmUlWxUulI8+bpXvFQiSaixBIx4sk4dpuJ6Bos48b4AicAW5VS2wFE5AHgQ0D6C2o+cJ31//PAvwoqYRbC8TDhWJgqT/dgjCC0hduYUjplFCUzFBGjrtvpRjeFiOB1lLKrcxuHVh7NzTfbufxyuP12XQVrtIkn40QSYSKJMIFYJzf/ZDJVE50sueRtajoSOMSB01aCXew47T6SKkE8ESOqwiRUgqRKkEQhaD+6gPUdRGy4bC6c9pIuY+20uTh2iZPjT3Hic5TmLWVwyZIm7r+/lJYWHUQ1XFLpSO+8o+dInjhxaPtpC7eRVGZmh6EynozvVGB32vc9wIkZ27wFXAL8GvgwUCYiE5VSzYURsTcdkd5FWX0uH7WdtUz2TTY5vwYoAt12OrUrNRrtmVLkspfQEQ3SEKrl2GOncdllcPfdcPHFcNhh+Thy/yiliCVjRJNhwvEQgVgHgbifWNLyrYri1ecmsfkdH1//Xh2TMma7j6CLiGT2ZvsiqZIZxjpJUsVJogCF2+7hAO90yl2Vg9pvNk46qYl7753J88/rnOp84Hbr4K4tW3Ra1VDGgWv9tfkRZj9lPBnfgfBN4HYR+SywGtgL9BqwEJGrgasBpkyZwsqVK0dMoHA8jELRJm1EghFq1tcAutXe8l7LsB/cfOL3+0f0WgyHYpYNCiJf3nRboQjFQrTaWnuuiEM8CirDw1iCoqW1iaB9Cx/9qIsnnzyBG24I8YtfvNn1Qo9E/NTU9D7WUFBKkSTZZQATGb0vG4JNhBKrxxqLCffcfhQzZnZy6vFvEcmoVKXiESINNUOWR+j9IlWqjb3sYR+C0+bCLo4hB6JNm+Zn8uQwjzzi57jj3hmynLl4+21dG7qkZODBckmVJBQPkYgkWPvi2i7vgGHgjCfjuxdIT3SYZi3rQim1D907QERKgY8opdoyd6SUuhO4E2DRokVq6dKlIyJwNBHlzdo3u1zONetrmLVA5xt2RjqZ6J3IzMqZI3LsobBy5UpG6loMl2KWDYYtX0F1O5qIsr52PZWeyh7Lw2E9j6y3vNdPCMeDJGwOjpw1n//5Hxs33ODijTeW8rGP6fU1NSuZNav3sQZDPBmnOdxAXXAPIiDYcNpcuG2uPhupj99XSV2dl5/8dg/eA3vn80YaaiiZnDvPd3gyxwjG/djEzhTPVKrc1Thtzv5/mEZNzUrOPdfNgw+6mTJl6bBndMpGR4ceYx7oOPCutl00BBtoeLeBJacsMfEpQ6B4ulXD5zVgrojMEhEXcDnwSPoGIlIt0vWU/g/wpwLL2IPOSGfOdT6Xj4ZAg4kkNECR6HaqclK2GsNuh5dQPEBLuJGLL4YTToBbb4WWluEfN6mStISbeK/1LeqCu/E5yyhzVlLqLKfE7u7T8HZ22Lh3+UQWLg6wcHEw53YjhcPmpNxVhcfupTa0i00tb1Ib2E00MbhJeJct05W21qwZGTnLy7XxfecdaO5noCKRTFAfqMfnHEa0lmH8GF+lVBy4BngK2AQ8qJR6V0R+KCIftDZbCmwWkS3AFOAnoyKsRVOwCbcje9kZm9hIqiSd0dwG2rB/UEy6PWFC79mAUqRKT8aSEW66SRd3uOWW4R2vM9rOlrYN7PJvpcTuptxVhV0GHln7wJ8mEPTbuPKrjcMTZJjYbQ7KnVX4nOU0huvY2Lqe3Z01hOM5LmYGixbpdK9nnhk5Gd1uXVBlyxbYvVuXxcxGZ7STpEqaCOdhMp7cziilHgcez1j2/bT//w78vdByZSOejNMeaaeipCLnNm6Hm3p/PZXuypzbGPYPikW3S8ugPkcaeqr05B7/DubMmcfnPy/ceSfMmgUNDTM4+eSBp98E4wFqA7vpiLXitZdS4Rp8mG/tHiePPFjJsgs7mHVItP8fFACb2ChzVqCUoj3aQlO4nqqSCUzyHITPmTs/y+HQxTpWruyusz0SOBy6gbVvny4YMnt279mf6v31xs2cB8ZNz3es4Y/6+53ByOPw0BZuI5oojheHwR+gupYAACAASURBVOD16KCcZI5J233OUtqjLbRHW/nCF3Rk9G23wT33zOS66/RMPX1NjRdNRNjVuZ0tbRuIJEJUuibisg/tRb/i9xOx2eEzXxi1ZIaciAg+ZxmVJRMIxP1sadvA1vaN+GMdqMwalRbLlulc6zfeGGnZdD5wZ2fv6Qkj8Qht4Ta8zhEYeN7PGFc937FEU7AJl73vaWBER5XQFm5jsm9ygSQzGHKTqnYVCGTP/YXu0pN168sRcVj1jvuemSeejNMYqqM+tBe72Cl3Vg0rze69d9yserqcT17ZTPXk+JD3M1SSClRSu26V0p/U/+l/9SmWYqOUlmCIfS0bcdu8THZPo8yp05Tilvgnn6wjkp95pntKxJEkWz5wW7hXDJ9hiBjjOwokkglaQi2Ul3SHjSpF10OWjs/po9Zfa4yvoWiorOxd7Sodh82JSgR57e0OotGe7uLMmXmSKklrpIl9gV0okpQ5K4adXqcU3PXraqomxPnop/MQ8TVAIhFtrES0a93h0I0VhxPsNrA7wGHvuS71QcAmHux2D9FEhGBiC05nCQd6p9PUpPfv9erqYc8+CzfeWJga2un5wFOnKprstfhcJtAqHxjjOwoEYgGUUj1eMoGALmCQmgs0hcvuoiXUQjAWNK4eQ1EwkFSXUkc5k2fV4/ZUEQp2WwkRmDlT5+r6Yx3s8e8gmgzjc5Rhz9NMXmtXlvLuei9fvbEejzeHfzxPJJP62VVKl9Q86CD9dzgTF5RQQhklRBNRdnVuJZqMoJS+dsuWwfPPw3vvDW6Go+GQGgfeuitAk0Q4Yo4XTKzVsDFjvqNAS6gFp71nFEN9g36A/f7e2ztsDlpChWvBGwx9kV7tKhciwpJT4syd78frVYgoXC5trP54V4J1u7awrWMTNrHrCOY8Gd5YDJb/tpoZsyOcc1F7XvaZjWhU58YGAjBpEhx6KMyZo121+ZoxyGV3MdE3ASVxOkI6TerMM/X+RzLqORsigLeJSNjB++/3nL3JMDSM8S0wqbl7PY5un104DJ1Wkntra+/f+Jw+6vx1po6qoWioqtJu1r7wuEq48baNfPenzXz60zv4xa+i/PC2WrZtFb79hYOJtU3uN+5hsDz+cCX7dru48iuN2PPs10sqbWzbLZt+8MFwxBFw4IEjO1G93Sbs7dA1VSZMgOOOK7zxTSQTNEcaqC73YbPB++9nf1cZBo4xvgUmEAuQSCZ65Mg1NsKbr/h48IHZvPCsr1cepd1mJ5FM9FmUw2AoJKWlufNA0yl3VzB70VY+etn7TD7mTY5YvI8f/2YPTfUlfPOqadTuGVy1p77wd9q4966JLDg+wKIl+euaxWJWL9evx7vnzdM93aqq4c1aNFAcdjtNweaunOBly7Tbeffufn6YR/zxDrCGykpKdBWsHTv0tckRnG3oB2N8C0xbuK2H4Y1E4Gufn8qvfnwg9997CL/+yYFc+bGpvdIxXHYXDcGMorQGwyhRUqLTiPpKGwKd1+pxeImrOGXOCnzOMo5ZGOZnv99NIGDnG1dNZ8fW/PR+/7ZiAp0dNq76atOwg5GU0q7V9nbdyJg+HebP15MQjER5x76w2cBpc9IQ2gfAWWfp5c8+WzgZGkN1uOzd3Xu7XUe9x+J9Dz8YcmOMbwFRStEYaOwROPWfR31sfc9DOGRDKSEcsvHOeg+rn+kZUeh1emkJthBL5GkWbINhGIjoXmA43P+2JXZ3rzlwDz0iwi1/1F23b31xOpvfHZ7ftr7Wwb8eqOSsCzqYc+jgSjemE4/rXm5np+7dHzJX93InTBi5whb9IQIeeyktkUYiiTAHH6x734UyvpFEGH+sA7e9Z3i7lQlpGCLG+BaQUDxEJBHBYQWXKAXrXishEu6pwpGw8O7bPQsLpHIe28MjF0RiMAyG8or+e759MXNOlF/ctZvS0gQ3/Pc03nojR+7SAFjx+2pE4L++NLSCGuGw7uVGozpi+fDDYcYMKPUVJqWnL0TA6RREOWgI6mn8li2D11/PT+3s/miPtGIb7YswDjHGt4B0RDp61KUNBGHazAhOV89BE1eJYvbc3q13r9Nr5tA0FA0etzYMwxnzO3BajFvv2s2kKTG++5WpvPLC4HNIt2ws4fkny7nkk61MmjL4ghqJhHajH3KIdi1PmtRzzuJiwOsFlyqjOdJANBFh2TLtDl+1amSPq5SiKVyHx25ye/ONMb4FpD5Qj8fZ3bpvbIATTg5QWpYA9BvM4Uwyb36I+QsCvX5f4ighEA0Qig2sGLvBMJLY7Tq1pr+o5/6YOCnBrXfuZuacKD/81kE8/1TZgH+rC2pMonJCnI99ZvDdwEhEn8fs2drNXKwdPJ9Pz0tsExuN4Trmz9dR1iMd9RyI+4kmIzgGOQ2ioX+M8S0QkXiEUCzUlVoRieixJbcbolEbRx0boqw8wozZUX76u710dGSPJrXZbLSGTIy/oTioqspPwE15ZZKf/X4P848J8b/fO4DHHs494Ug6L6/2sWGdl09f3YzXN/gueDg8emO5gyFVTcznKKMxVEcsGeWss+DFF3PPMpUPWsONOKTI3ADjBGN8C0RnpLNHrdrWVt3KfvMVL/4OO5d8spUzzqxlV42LREx0tGWWh6rUVUqdvy5n8XWDoZDkKjE5FHylSX78672ccHKA3/6/KTx0d1Wf28fjsPy3k5g+M8J5Hxp8LEQioXu9+SqKMZKkZhayiQ2b2GgO17NsmW48jNQcv/FknNZIE16HcTmPBGNA7cYHjcHGrsIaiQQ0NGhX0qqny/D6Eiw8KcjCRY3Eojbeet2L3Q4dWd4nDpuDWDKGP5qlFJbBUGBcLu29ieUpCL/ErfjeLfs4/ZwOlv92En/+3cScY8pP/LOCPTtdXHFt05AKagSDMHly8bqa03G5usfWfY4yGkK1HH1slPLykYt67oy2o8g+81oiAa+84OOhvx3Ck0/YhhV4t79ijG8BiCVidEQ6cDt0OkVHR2oiBeGlVaUsWerH5VIccWQrbk+S19b4cLt18fpsLx6n3UlDwOT8GoqDiRMHlnI0UBwOuP6HdVxwSRt/WzGR3/18cq8hmIDfxj13TeTohUFOPLV3fMRAUEqnS40FbDYdFBaL6d6vCHQkGli6FJ57LvukLMOlKVyH2947qTmRgO9cO5Wf3nggf7tvLp//rINzzx1e5Pv+iDG+BSC9MpVSejJyjwfWvewl4Ldz2tl+kgpEFMceH+TVtbqEWyyW/aXmc/poDjUTTxZ+qjSDIROfL/9Vjux2uPaGBi79TAuP/qOSW246oIeBeejuKtpbHVz11cYh9VzDYSgr0wZtrOAr7TayXkcZjaF9nHFWnLY2WLcuv8cKx0ME4p1Z51J+fa2P997xEAnr6ZgCAeGVV+CJJ/Irw3jHGN8C0BxqpsShlTgQ1A++06ldzqXlCY49IUA0ot1fxy0O0FDrZOd2FyI62T8TEUEpZXJ+DUWB262NZb57PiJwxbVNfO7LjTz/ZDk/uv4g1jxXyl2/qubvf63ijPM6mHv40EKtIxGdUjSW8Hm73ft2saNEcfiiRlyu/Ec9t0VbsOUwD1s3lxDOqE0QCMD69fmVYbxjjO8Ik5q7NzXe29SoDW8kLLy8upSTz/DjdOqHym6HI4/TLrTX1mrXc64keo/TQ72/vlCnYTDkRERXgMqn6zmdyz7bype+Wc8rL5Tyk/85kH/cW0U8LtTXOoZk8BMJ7dr2jbE4IldGJ9TnKKdT9nDSkiTPPJM/70NSJWkO1+N1lGZdP2depJe3weeDBQvyc/z9BWN8R5hUYJSIEI3qKjoejzauoaCN08/WXVul9Ath4qQ4s+ZGeG2NTxvpSPZUDrfDTUe0g3B8hN54BsMgKC8f2TG/Aw6K43QlSSYFXdRQ2L7FzetrB29Bg0Hd6x0LUc7puDJSbe1iR6kki0/rYO9e2Lw5P8cJxv1EE9Gc0zwGgzZUUnA6k4gofD7FiSfC+efn5/j7C2NM/cYezcHmrrl7U+lFIvDCM6VUVMY5ZmGQWEyPPdlt+oWw6KQA7673EPDr2xPIEU9iw+T8GoqDVMrRSGXAbd1cQjzWuwzrti2DH7RVSucnjzWcTv3uSKZdY5+jjHkn7EBE5c313BxuyDnVYyQsrPhdNbPnhbnxZ7Vc9okt/GlFnKeeKswMT+MJY3xHkKRK0hxqxuv09kgvCoeEl18o5eQz/dgdundbVQVYxeqPWRQgkRDefNVLSUlu17PP5TM5v4aiIDXLzXCrXeXikEMjlLh76nmJWzFn3uAOmAq0KrbykQNBRL8/4mlpXXabg9KqCEctiOUl5SiWjNEaac5ZTvKf91fSUOfkC19v5KTTAnz0Y9s47/ykMbxDoCiNr4icLCJPi8gWEdkuIjUisn0AvztPRDaLyFYRuSHL+oNF5HkReVNE3haRC0bmDDSBaICkSmITG52dumKVzQavvugjEu52OSeTOpIRtPtuzmEhfKUJXl3jo6QE/P7seZROu5NIIkIgNrRUC0PhWbNmDWeffTbz5s1j9uzZAEeNRd3ORkVF/vJ9M1m0JMBhR4Zwe7Sr0+1JctiRIRYtGZzuj8VAq3S83t7XuNRRxrEn17NxI+zdO7z9d0Z1EGe23N7WZjt/WzGRk073c8wiU+J2uBRrYbXlwNeBN4ABjSSJiB34HXA2sAd4TUQeUUptTNvsu8CDSqk7RGQ+8DgwM5+Cp9Maau2awai+XkeFgo5yrpoY58hjQyilW7Rey23n8ehexMLFQV5b4+ty4wWD+uWWidPmpCnYRKkre3CEobi44oor+OUvf8nChQux2+1UV1dvBJb19Zti1O1sjETKUQq7HX7y2728vtbHti0lzJkXYdGSwKB6XImEdt2OtUCrdLxe7UFLx25zsPDUZv5y+3SefRY+85mh778xXJuz13v3HyYSjQhXfqVx6AcwdFGUPV+gXSn1hFKqQSnVnPr085sTgK1Kqe1KqSjwAPChjG0UUG79XwHsy6/YaQdSisagnrs3GNT1V10uCAaE19b6OPWsTux23RIvL+8O/nC59PjvwsUBWpsdbNtSgsulx4uz4XP5aAg0kEiaDPexQEVFBeeffz6TJ09m4sSJAImxptu5yHe1q0zsdjjx1ACfuKKFE08dnOEF3YCtrh57gVbpuFzZK3LNnulk+uwQzzwz9NZPKB4kFA9kHe/d/r6Lpx6p4IMfa2PqwWZO8XxQrGr4vIjcIiInichxqU8/v5kK7E77vsdals7NwKdEZA+6Z3Bt3iTOIBgLEk/GsdvsNDV112Z95YVSohEbpy3TUdDRaO8ebWUlHHGsdqe9+qJ2PXd0ZI8mtYkNpRSd0SwJwYai44wzzuBb3/oWL730Eut0ZQTvWNPtvhjJlKPhkkyOzUCrdJzO7N4Fh83J8ae28PrruRvq/dEWacEmvVs0SsFdv5qEryzJJ64c2nzJht4Uq9v5ROvvorRlCjhzmPv9OLBCKfULETkJ+KuIHKmU6lG8TkSuBq4GmDJlCitXrhz0gWLJGLFEjFbaCIV1qz3ih+cfPZaJE8PMOXAjkQZwJKClBtp2QiQYoWZ9DYkETHDC3HnVvLLSzkcuqsGegJr12SMKkyrJy7zcVb5ypPD7/UO6FoWgmGWDbvmeeuopAJ7pDk2dBtxKEem2QhGKhWi1Df4tnkyCPQyRjCFBFY8QaagZ9P7yhUqC2wZ7N/Zel3ruipVM+Zwh6/pm9ICXHFvGw3+ZykN/38TZywZXA0AB4XgQl9iI0DPC8/VXq3nzVR9XfmETrvAuImmNK7uK8/ora7HZxkCB7CKjKI2vUuqMIfxsLzA97fs0a1k6VwDnWcd4SUTcQDXQYxRFKXUncCfAokWL1NKlSwctzPra9TjtTlqbnXTWgrcc/J023lxXzYUfbcdzwCzicVBxmHO4/k3N+hpmLZhFIgHvvAMnnBbnvuUTCLvm4HQlcVfAtGnZj9caamXBAQu6KmmNBCtXrmQo16IQFLNs0C3fm2++2WO5iGxRSvVneAuq29FElPW166n0DL7wsVLw7rvg9PR070YaaiiZPGvQ+8sXHR16zt6yLFMFp567YiVTvh079DCWO6OtfcQkmDApwuvrZnP1VYcP6hid0Xa2dmyi0jWhx/J4HP68YiZTD45y8WftOBw9r5N/31YWn7wE71iq01kkFKXbWUQqROQ2EXnd+vxCRPqb4PM1YK6IzBIRF3A58EjGNruAs6xjHA64gbxHD4RiISKJCHZxdqUXAby0qpRYzMZpVpRzV4pRBna7nth7wQkBlBLWvayrXbW29h3Q0hZuy/epGPJMe3s71113HYsWLWLRokUA08aSbveHWOlyxeR6HqsVrXLh82UfVxeBxaf7eWmNk2BwcGO/TeF63LbenrPHH65kz04XV321sde8x7EYiM3k9w6VojS+wJ+ATuBj1qcD+HNfP1BKxYFrgKeATejIz3dF5Ici8kFrs28AV4nIW8D9wGfVCCTJdkQ6EAS/v3vOUIDVT5cx+cAYhx2p30yJhDay2aiogINnh6moivPqGj3RQjKpg0ay4XP5qAuYnN9i5/Of/zxlZWU8+OCDPPjgg6Cj+ceMbg+EysqRmWVnqIzVila5cLtzN8JPWRoiGrHxzOqBTzkaTURpj7ZSYu85OXNnh42/3jmRBYuyzxwViejCQIahUZRuZ2COUuojad9/ICL9lu1WSj2ODjZJX/b9tP83AifnTcocNAQa8Lq87NzTPWtKZ7uNda94+fAnWhHpfng8Hjh5+ck0hZr0ghf1n4meau5YtIZFJ+mUo0RCt2z9/uwteJfdRUuwhWAsiM81Tpr445Bt27bxj3/8I31RLTC7v98Vi24PhPRqV8UwV24yOXamDhwIfRUIOXphEF9pgiefjnHRudnn4s2kM9amC3ZmbHv/8gn4O2xc/fWGrPcxVbfAMDSK9dKFROSU1BcRORkYE1ndkXhERzpHXAQC3cZ3zcpSEgnhtGXa5RyN6vEnu51uw5tGc6hJl5pcEqCj3c6WjW48ntzVrgAcdgfNQRONWMx4PB5efPHF9EU+xohuDxS7Xet2tprkhSYc1l6ksVjRKhepzImsUc8OOOGUAK++UEZraGCznjWGavE4ejbY9+128siDVZzzwQ5mz8t9I6VYLcgYoFgv3ZeA34nIDhHZCdwOfHGUZRoQnZFOEGhuoccYyeqnyzhwarRrCrRc473plJfDEQsC2GyK19b4cDj0Cy3XeJrP6aM+UE9SJbNvYBh17rjjDr785S8zc+ZMZsyYATCDMaLbg6GqauRKTQ6GSETn9o4nbDbtes7l2l+y1E9nu5PnX27udxgqGA8Qjodw2nq2Tv7vN9U4nIr/+mLvjgF0FywxQc5DpyiNr1JqvVLqGOBo4Cil1LFKqbdGW66B0BhsxJ5009qiq9EAtLXaWf+6l9PO7uzhvvF4su8jRWUleLxJDjsqzKtrdMs05XrOht1mJ5FM6AaAoShZsGABb731Fm+//TYbNmwA2DhWdHswFENw03ioaJULb46gK9AFepzOJGtWefDHOvrcT1ukGYet53RJb7/hYe3KMi77bAsTqrMX74lEskeOGwZOUY35isinlFL3iMh1GcsBUErdNiqCDZBYIkZHpIN4QA8wpQztmudKSSakq5Zz6qXQX3R+yjifcHKAFb+vprnJTnlFgubm3K35EkcJDYEGKtz9BdAaCsnTTz/N0qVLue22Xio8RUSuK3bdHiypSm2xWLebtNAEg3DAAeNzXLLUBy05Rpi8PsWCE4K88cJE9gXeZ56zPOvYbyKZoDFcR6mjvGtZMgl3/moSk6bEuOQTufO843FtfCOm0uSQKTa1TLVRy3J8ippALEAyCQ0N0qNXu/qZMqbNiDJrrh47iUR0rzb1PHgcvbvA1Z5qXC794jruRB1p+MZaPcdvOJx7PM3r9NISaiGWMCXgiolQSA/rdnZ29vign8Gi1+2hMJrVrpTSn/EUaJVOf2PYS04PULe3hPe2JAjEs3vCAvFOlDXxS4pnHy9n63tuPn9NU69ZpDLJzDM2DI6i6vkqpf5o/f3BaMsyFJqCTcQjLmKxbpdzS5OdDes8XP65li5jG4/r8dwU67+oA7m3vbmN98vep7yknCXTlwA6WOSggyNMnBTn1bU+zvmgdiMFAtkfQBEBgfZwO9W+4h3s2rNH9973lwf4gx/UGUE33XRTj+U333xz7VjV9/4oLR25iRb6I1UzfTwFWqWTmts3Fyee6kdkMm++OJkjDtvDIRXze23TGKrrkV4UDgkrfl/NoUeEOP2c3ENXSSuK3dTVGB7F1vMFQET+V0TKRcQpIs+KSKOIfGq05eqLRDJBS6iFjmZvD4Py4nNlJJPSVVgjlX6R2qa2s7YrKEIQfvnyL/ntq7/t+n15uXYFHb8kwLqXvcTjWun7qt/qc/qo9dfm/RzzRSgEr763h8bWIqrEUCCuv/56Ojo6iMVinHXWWQDHFLtuD5XUDF3JUYj/G+tTB/ZHqlGRzNG4mVCd4PCjwry2uorOaDv+WE9jGk1E6Iy14U4zvg/9dQLNjQ6+8PXGPl31sahuWBVDGtlYpiiNL3COUqoDuBDYARwCfGtUJeqHQCxAKKQIBKRHi3DV06XMmB1h5hztJ47FdACIw6HTki66/yJuWXsLoHutl86/lHW169jWsg3QRloEjj85QDBgZ+NbHlwuHXSVK9rRZXcRjAUJxnJU5Bhl3tvRyr7wNrbX7X9pUf/5z38oLy/n0UcfZebMmQDvUOS6PVRGq9rVeA60SsdXqg1hLk5a6mfre246G8upD/asRtoRbesxDtxY7+Dvd1dx2rJO5h/T9w2LRnt67gxDo6jczmmk5PoA8JBSqn0gyeKjSUuohY42Z49Sa431Dt5d7+XTX+gO109PfVi1cxWd0c4uFzPAxYddzC9f/iUPbXyIG065AYdD9yCOWBDE4VC8ttbH0Qv1PMDBYO6HwCY2WkIteJ3ekTjdIdMZiLF+13amVleyq6mBROIg7Pbivrf5JG61mB577DEuvfRS/vSnPyUookbwAbceQH2gZ1H+ak81a65YM6T9VVRAczMUsgJhKtCq0K+MHsVyLIZz7bIV30nfX6kPGoO53b9LTvez/DeTWLdmEmUXbyMQ8+NzlurpTkO1eO3d5fX+csdEkkn4/DX9R1Ap1X+mhqF/iuahz+BREXkPWAg8KyKTgKL1USZVkvqOJoIdHrxpre0Xn9XKnSqsAVpxU+PB/978byZ6JrJ42uKu9dXeas6adRb/eu9fRBO6WVtVBXZHkiMXhHjNSjlyOKC9jxz6Ulcpe9r3EIj2Lgs3mry5bTc2exKvw0M0EaWpozh75yPFhRdeyGGHHcYbb7yRcjs7KCLdzjS8kL0IzEAp9Eu6UIFWSZWktrOWtbvXcu+Ge3l93+tZr1NTqIkly5dw/zv3A7CzbSdLli/p9fnXe/8CYFPjpq5lufaXwu3u26U/9eAYB8+K8NLKUkrsbhpCeornYDxAJBnuSjHasrGEZx6r4MMfb+OAqQOrC7q/xGqMJEXZ81VK3SAi/wu0K6USIhKg9+ThRUMwFqSlLYENe4+k81XPlDF7bpjpM3XkcTKpx8Dcbl3/+fkdz3P5kZfjsPW8DZfOv5TNTZvZ27GXWVWz8Hqtcd+TA9z160k01DmonhKntRWmTs2eSmG32fE4PWxt3cqRk47Ebhv96ucN7e28X1vPQVV65hSH3c6upmamVI1z/2AaP/vZz7j++uupqKjArt0kSYpYt4eLw2GlpOShhtfHnz6Z1mhPg1Tlqub+s7t7luFwfgOt2sPt7GjbgcPm4IjJRxBNRLn0oUvZ0baDcLy7zfTZYz6bcx9nzzmbGRUzAJ2NcPacs3ttM61cT1dWXlLetf6Bdx7oUzaXq/+AtiVL/Tx49wTigTLaPI0E4wFawo04RF8gpXRqUUVVnMs+10f5PItYTL+/zGQKw6eojK+InKmUek5ELklblr7Jw4WXqn9aQ220NNmpSGvl19c6eG+Dh8/+d7cbJxzuTjH6z7b/EEvG+OChH+y1v5MPPpknP/Vk17m73drALlyije9ra3x84CPt2vUc0u6nbHicHtrCbezp2MOMyhl5PefBEk/GeX3bNnzOsq65Pyu8PmrqGll0yPQB1aAdy6xbt46lS5fy8MO9VLgcWEKR6nY+qKqCfXXDr/WcaXhTy1buewy33cPiKWcSjUKifBvr6zpx2py47C6cdiceh4cppVMAHWthExtLVyzN6ta9YN4FbKjfwI62HbSGdWTj2bPP5vYLbsdldzF3wlwWT13MrKpZzKycyazKWUz2TWbFWyuyyv2Dpd3B7JN8k3p8z2Rq+dSu9QMxvv1x0ul+HvjzRF5Z42PJOR3UB/fQEW2j1KnrAKx5vpR33vRy7f/U4yvtPzIuGh1/FcNGi6IyvsDpwHPARVnWKYrwBaWUYkdDI7akt0cxgdXP6NTN08/uLkcVj+sxMIAPzP0Ale5Kjpp8VK99pvLuwvEwsUSMspIyysvB4Ygy5aAYr1rG126HjvbcxhegoqSCvZ17qSipGNL8rPlie9Nu6pviHFDVPc7kKbHT3ByjLeinyjcuU127eOstXcTq3//+d+aqSnRgYdHpdjpv1L7BwgMXDum3FRXQYA2TlFcMrSRhXyVTf/bmdVS4qrj3jJdxOuH3b93GMzXP9NhmatlUnvuv5wD44mNfZO3utVn31RRqYnf7bpw2J2fPPlsb16pZzJ0wt2ubW8+5dfAnMALY7Xq8N33mtEzmHh6henKMl1aWsuyCUtqjLaD0OyYaFZb/ppoZsyOc98GB1YFOJMZ/IFuhKCrjq5S6yfr7udGWZaCE4iH21Ufxuntq5Oqny5h7eJgDp/UsdpEaK/E4PSybvSznfgPRAGfdfRaXHXkZX1/8dSoq9Mvr+CUBnn60nGhEcLsVLS1w4IG5exQiQnlJOVtbtnL0AUfjshc+9Jb55AAAIABJREFU8bEj0sGm3XWUOib0evE6xMneluZxb3w/9zmt0n/+c8/ZA1esWLFDKfX50ZApG1N8U3qN+9rExraWbUM2vjYbuJxQNRX27u2eUGSgbGvfxK83fC/n+jtPe5wkSYJB/SxcO+VaLjvyMqKJKLFkjFgiRom9OyrpI4d/hBOmnsCvXv5V1v394cI/DFy4NKo91VkDrobKQPbn8egAs1zX02aDxacGeObxciJhweMsxW417h/5WyW1e138+Dd7sA/CEpj83vxQVMY3hYj8FPhfpVSb9b0K+IZS6rujK1lv6lvbCfiFyrRnYt8eJ+9vcnPlV7pdztGoDrRyOuGRzY9Q21nLVQuv6lFdJh2fy8fRU47m4U0Pc+0J1+L16lt1wikBHv17JRve9LBwcZB4XLuz+wpscdldROIRalprmDdxXkFdvPFknPcathFoK6WqrPdxvU4fO5samT/t4JzXYjxx4403cv3111PZHRFkF5EfF4tu132zjmgiyvra9V2ekmAs2FWFbdXOVcydMJeDyg4a3I4FJk/WrtIdO7qfhb4IxQPcs+V2/rnjL5Q7c3ttDi6bg1LQ2amHdSa7DuOw6sNybn/hvAsBchrfoTLUqOb+9vfpez/N3vhenv3Ms72eXZ9PN8r7CoA6aamfR/9RyfrXvJx4qh4kbmu1c9/yCRy/JMCikwYW9JhK4RqvhUsKTbG+7c5PGV4ApVQrcMEoypOTzXsb8GSk86x+WrtWT02Lck6fxWjF+hX8Z/t/+jU2HzviYzQEGli9czUlJVrxj1gQxFWS7Ip6ttn0S6c/ykrKaA210hgsbDHWvR17aWqJ4ZSSrIFhHreNltYk/ujAJ/8eyzzxxBPphhcgQZHqdgqv04uIEIlH+M6z3+FDD3yIp7Y+NaR9VVbCIXN1g7Gv/N/XG1Zz9aoP8I+aP3He9I9y19InqHL17kWmlo3HqQNTnDTxJPZ27mVz8+Ze69zu/oOujl4YxOtLsHZV95DPvXdNJBy2ceVXB/4+SE2DasgPxWp87SLS5dwQEQ9QdM4OfzhMbWOYyvKeT/zqp8s47KgQUw7sDttPpRhta93Gu43vctG8bMPaPTl9xulUe6t58N0HAf1yEVEcszDEa2u18XW7+57jN50KdwU1rTUFK77RGelkd9s+As0VlJZm38bpBBV3Utc+9HSWsUQikSDSc649oQh1OxsljhLuveReZlTM4CtPfoXvPfc9QrHBhzGX+mDePP1/IEcmXEukEZ+zjNtOup+vHPVDypwV3H/2Gp78wOYen1Sk81ACgbK5hIfjJh4pTpxwIoLwbM2zvda5XP0HsTmd2mP28mofiQTs3O7isYcruODD7cyYPfBJl2Mxcj7HhsFTrMb3XnR+7xUicgXwNPCXUZapF3saOlH0DCDZvcPJ9vfdnJ7W600q3UP1eHRur01sXDC3/86O0+7kksMuYdXOVTQGGikv10Fbx58cYO8uF3t3OXE6da96IHOn2m12ShwlbG3ZSiKZfaqwfJFIJtjeup1YsBSQPsvVldi97GltGnGZioFPfvKTnHXWWSxfvpzly5cDzKMIdTsXMypncN9H7uOq467ioY0P8ZEHPzIkr4XbDXPm6PHDzk5IqAT/qrmbp3b/HYCzp13C7ac8zPwJx/W7r3hcG6HBBgKtuWINm6/ZzJOnPMnmazaz+ZrNeXcd54MJrgksOGABz27vbXxTrvt+U45O99Pe6mDTBjf/95tq3O4kn7p68BXmTHGN/FGUY75KqZ+LyFtAKiLpR0qpofm5RpAd9a2UZQy2rH66DBHFqcu6X0jhUCrpX/HolkdZPG0xk32TB3SMTx79SZbNXka1t7qrnOTxS3R34bW1PqYerL3zgcDAAiG8Ti+t4dYRTz+q9dcSjIZpbazq96Xoctro6FD4o/5xPxXit7/9bY455hieeaYrGrdWKfW/oynTYHHZXXxzyTc5efrJvLL3FUpdQ+sOuVwwezY8t+Edblv9fbb73+X0Ay/g3OkfRURwyMDmIgwG4aCDxnet4TNnncmvX/k1LaEWJngmdC1P1YmPx/seQ19wQhC7PcnPvnMgTQ1OPndNI5VVA2/sKjOZQt4p1p4vwCbgSaXUN4EXRKToRhsSGdNxgU4xOmJBiOrJ3S7nVIpRR6SD2VWzufiwiwd8jANKD+CYA45BRHA6dcuzekqMaTOivLqm2/Xc10QLmVSWVLKvcx8dkb4n2h4qgWiA3e27UaHKPtMgUpSUQDjgoiHYMCLyFBuHH3445513HrfeeiuAvxh1eyCcNP0kvrb4awBsqN/ANY9fQ0togGMggD/q5+f/P3tnHh9Fef/x9zOz92azSUgINwTkTrghHAoo4oFWq/XWqvWuVn+29ULrfdV6tVV7iFosWLQIVq0HyimHlUMRuc9wJ0DOTfbeeX5/DFlyZzfZhIDz9rUvzM7MM08mz8x3nu/z/X6+K57hruWXUhI+xP/1fZn7BsdX1rhyxuc+ud/ZuHzg5Sy/YXk1w1uJ06m7hOsjEoFnpnZESsGRQ2YQkjVfO4jE4WgKGsUUEk6bNL5CiJuB94G/H/2qM/Cf49ej2MjbYWH3Tivjz6zthrPZ9DXX13/yOhf2jU/QqMhXxMMLH2bl/pVRofqR4yr4YY0dv09gsegz34ZuwKoIIXBZXWwr3Jbwur+a1NhevB2b6qCgQESlNBtCVUHVHBSUFhPWYpO3O1GZNm0al1xyCbfeemvlV2ZOgLHdGDtLdrJk9xIumHVBvTm0NdlwaAMz1s3giuwrmPfzT7ls6BQ8HlFvwZC6OJkDraritrlJsdUd8e1w1F9kBWD1Cieb19vRtKOWUwq2brSzekXsfvqaxRTGvTmOvq/25bwV52F72oZ4XNDhhQ4xt2fQRo0vcAcwDigDkFJuA2Lz0x5HvvrShaJITj3j2HpvpRybUEPkl+c3qV2n2ckXO75g1vpZOJ362/7IsRWEQgprV+nWrbLQQqxU5vvuKtkVLWmYCPI9+QRCAYJeO+GwLi8YGwKfX+IJxBC6fQLz2muvsXz5cpKPPckCnABjuzEu7Hshsy+djcvq4oYPb+D5Fc9HtcmrctBzMKpjnNsll3nXzOPRCY+SbE0mPV13Q3u9+sM+FkKhH4/i0uoDq7n+P9fXukcae/HYvsVKwF99yhrwC3Zsjd2HXLOYQl2603XpghvUT1s1vgEpZfT2E0KY0BWuGkQIcY4QYosQYrsQ4oE6tr8shFh79LNVCFFSVztNQUpY8qWLnGE+0tKP+XP8fj3FaMXeFUycPpHVB1bH3bbVZOWCfhfw5Y4v8VOEEDBwiA+bXYtGPVss8bmeQU8/KvQWJiz9yBvysqd0D0mWZA4ejC84w2SCYIWNQxUnt+vZarViqf20bNNjO1b6pfdj7mVzuWzgZbzx7RvM3jg7ui0iI/zju38w5V9TePKrJyn164pK3dzdqrXhdkPv3rrx9TUSSF25zvljUlz6et/XLN2ztNp3jRnfU/oGsNqqDzGrTdKrTwxRmkcx1nsTT5sMuAKWCCEeBOxCiMnA7UAtXb6qCCFU4DVgMrAPWCWE+EhKubFyHynlr6vsfycwNFEd3rnVyv49Fi6+uroFlFJ/OHy89mOSrckMyhzUpPYvHXAp//z+n3y09UNOT/0Ffr9kWK6XVcudSKnfGKWlDUvN1YXb5iavOA+XxYXd3PRQRk1qbC/ajs1so9yjEAzGZ3xtNvCW2yj2lRCKhDCrsQXbnGhMmDCBZ555Bp/Px5dffgnQC/hLQ8cc77EdD3aznSdOf4Kze51NbpfcOsvspdpSGwysczj0VKRdu/S61fWlt3i9emGRH8s65NAOQ0mzp7Fg54Jq2RJms37Pa1rdRVZGjK2gX7aPzevtBPwCq03SL9vHiLGxVTwLhfTnS+xeLINYaKsz3/uBw8APwK3Ap0BjCkCjgO1Syp1HZ83v0nC1mCuBWQnoKwBfzU9CUau7nLWjEYLS5GXBrgWcc8o5TZZ37NOuD0MyhzB742ySkyWBgO56PpRvZvdOS/QB1NhsoSYmxYTFZGFH8Y4G9XMbo6C8AG/Qi93kID+fmNZ6q6IooGmCYFC0WCBYW+C5554jIyODnJwc/v73vwOU0sbHdlMY120cJsVUp3uyslhBQ1iteiqSwwFldQyHypWSH1NRd1VRmdhjIkt2L6kVq+Fw1B/zoarw9Cv7mfr0QX5+ayFTnz7I06/sj/klveZ6r0FiaHPvMkff8jdIKfsB0+I4tDOwt8rP+4Dces7RHchCL+LQbKTU13uHjPTiTjlmwIIB3Y22KG8B3pA3JmGNhrh28LWsObgGxeoH7NE311UrnPToFcRkgpKS+BPhHWYHJb4S9pftp6u7a9z98oV87Cndg9vmxuPRXwDqq6caiegBINu3WDmlb4ARYyuqPQRkSHc9t3O0i7sfbZ1IJMLAgQPZvHkzN998MwBCiCOy8UX34za2jydmM2Rl6XrQhYW6Aah8yfyxBFrVZFLWJOZumsvqA6sZ03VM9Huns+F0Q1WF3NMqyD0t/vrekUj1Z0pEi6CgoFH9ZT3TmRl32z9m2pzxPVq/d4sQopuUck8LneYK4H0pZZ3B9kKIW4BbADIzM1m8eHGdjYS9flRNsuHbFA7u78PPfraVwKH90e1aBEJe+Pfmf5NhzSDtUBq7Du+qt1MBb4Bda+vfPoABDHAPoHhrPmY/WAX0yMrgm0UqF5y9CxV9lrDrMLpuUjxIOCwPs820rV7Zy/Ly8jqvhT/sR0pJiSjBHwAbEKhj6TYSgcceHsHWLW4CARWrNUKfvqU89uRqVBUsGpTuhPK9YfLN+Yg4fon6+tZWKC8vZ+nSpbRr14733nuPzMwWe1A1e2xLJL6Qj2IlziCCOGhonNfRIRxh8OUfW1KREQj7YFcCVrYbu++ON1X71znSmSHuIeTvyGdX4bE+RyKgBCDQAuJ1pggc3gZHjj4WVhevRkPjd/1+x0jHSFwuV/Rebcv3YFujzRnfo6QCG4QQK4Hoq5qUsnbx22PsB6pO27oc/a4urkCPqK4TKeXrwOsAI0aMkBMnTqxzv09WbsYfCPH16s6oqmTCBTas7qzo9rIy6N0fXuz3IntK99CrU68Guq8/kLKGZDW4j5SSNQfXkBToS8TrYtT4EHNmphJ29MKZpFFWBt1PaVoQSjASJBgOkpOZU+ea6+LFi6l5LQrKC9hVsos0exoeD2zfXv+s95ulTrZuTcXvP1oy0W9i69ZUftieTe5pFWgSKsqha+9ieqX2It0ZexhrXX1rS1T2T1VVbrrpJkaNGoVT/yOdcnT9ts2M7ZqFFZrFsrq/bmyc10VREezefTR7QEKv/olZ743lvjue1Ozfe8Pfq7WP3w9btoAjwe5hTYOIH3oOPPbdnz7/E6m2VC4//XL2fL+HsaeOxWoyorHipa0a3/rrh9XPKqC3ECIL/cF0BXBVzZ2EEP3QjfvXzerhUaSEpfNdDBtdgct9zA1TGaRgsUB7S/uYFa0aY9ORTVw992ruz32U4ZarGDWugn+/nca33zg4bVJ5tNBCU4yvRbXgD/nZXbqbXqm9Gq1+5A/72V2yG7fVjZQ0GuHcUMpD7mkVKOKokk7YQX5FflzG90ThySefrPbzxx9/nA+82Mhhx2VsJ4JEltlLS9Nd0Tt3/rgCreqiLFBGRIuQatertTRWIaqpBALViylIKakIVXBh3wuPS3nSk4k2ZXyFEDbgNuAU9GCrN6WUMaXcSynDQohfAfMAFXhLSrlBCPEEsFpK+dHRXa8A3o1hnS0mtm20cyjfzLW3VX/ABAJ6CbVnlz5Lbpdczsg6IxGno396f/ql9+OjHbMZ3v8q+uf4cCZFWLXcyWmTyqOFFjIzm/ZwSrYlc7jiMCnWlAaNn5SSXcW7MKkmVEWlvEKPPm1IaeiUvgFMJkkodKxjNVMehIBI0Eq5qQh/2I/N1ECttBOIYDDIH//4R7Zv305OTg433ngjJj18tFxKuaShY4/X2E4ElVrJiZpdulzQv398Ef0nG+XBcsa9NY5bht3Cnbl3Avr1sFqJM7e+cUKh6sZXCMG0n0xrVnCmgU5bi3Z+GxiBbnjPpfEZQTWklJ9KKftIKXtJKZ8++t0jVR5OSCkfk1LWypNsKssXpmA2a4yZUD2QQdOgRO5l+vfT2Va4LVGnQwjBZQMuY9ORjewPbkCTMHy0l1UrnGiafuOFQrEVWqgPt83NzpKd+MP113w77D1Mqb80quubf7DxPMBT+vqJaAJFkeiprZK+A6unPFgsUFIKAhHNBT0ZePbZZ1m9ejU5OTl89tln/Pa3v43r+OMxttsqFsuP2/gmWZLIbp9dq8pRQxHPzaGqfH2lfOiPofZ2S9PWruAAKeU1Usq/A5cApx3vDjWEpsHXi90MH+PFmXTsTbBShHzh3k+AY8W7E8X5fc7HqlpZfPjfUanJ4kJTVLFGCD0/sqmYFBNmxcyOorrTjwLhAHkleSTb9AWmigr9fA0V9AZ489UMBHD7fQVMmOwBBJdeW1ztQWqxQLkHbKqT/IqmKYK1RXbv3s3MmTO59dZbef/991m6dGnjBxkY1MOkrElsOrKJA54D0e+SkhqWmYyXmsUUNh7eyKlvncqiXYsSd5IfMW3N+Ebf22J1Nx9PNn5vp+iw+aghOYZedFry320fM7zjcDond07oed02N+eccg6rDy8nomnHqhxVKbRQGH+1sGo4LU7Kg+XVbm7Q3c15JXmYFBMmRfdvFRQ0nvLx/Wo7Cz9L5tLrijj/Z2X89tECklwRFn5WPUKk0lWuhfT156bUi22LmKr4Ak2GWoFBM5mUNQmAhbuOZZQlOu0qGARn0jHhjjkb56AqKkM7Hhf9lpOOtmZ8Bwshyo5+PMCgyv8XQrQZ5YVIBP77X3jrz+0xmTRGjK0+zQwE4JC2me1F2/lJ3+bl9tbHA6c+wCdXfoqqKLhTI/QZ4I8aX7NZj36MVR+3Ptw2N/tK91XTki30FlLkK4q6m71ePaq7oUCrUAhefa49HToFufIXutvKYpWMn+xh+aIkvBXVF6eF0GfTilAo8be6SmKLsGPHDpKTk0lOTsblcrFu3bpKfeehbWlsG5wYZKVmkZWSVc31bLE0Xtc3HoJBcB99Nw6EA3y89WPO7HlmvQUeDOKjTRlfKaUqpUw++nFJKU1V/r9NaKxEInD22XDllbB1ow1NEzz9QKda5bnCqofs9tmc3evsFulHmj0Nu8WC0ykJBnXX8+b1NkpLjv1JK+LPp6+GIhScFifbi7YT1sJIJDtLdlaTBjx0qPFIyzkz09ibZ+X2+w5X05g9c0oZAb/C8kXVK+rZbLpUpsPsoKC8IKGFH44XCxYsoKysjLKyMjweD+FwmDJduum7tjK2DU4snjj9CR4efywxxGw+qqiXoNtFSrAdfalesGsBpYFSftb/Z4lp3KBtGd8Tgc8+g2++qVxTFWiaYPP6Y+W5wmH9DXRcj1HMuWxOnfU3E8Xa/LXc9NWZ7CjazqhxFUgp+PZ/x1zPe/fqxRaaczNaTVbCWpi8kjyCkSCqUKPuZp9PV9RqSEoyf7+Jf72ZxrjTPYwaV/1toP8gPx27BFnwSXXbYzbrbQtpJhAJ4AufHK5nA4NEMqrzKHqm9oz+LETj5QXjxXZ0vfeDzR/QIakDY7qMafgAg5gxjG+cfPdd7Rll1fJcgQBEbIepCDZz2hkDXZO7cshXwIKC2fTu78edGmblsmNVjmw2XZRg61Yob0Z33DY3hysOE9bCuKzHZqmHDzc865US/vJCexQFbvtN7cpJQsCkKWV8v8bO4XxTrWP9fn32XexrOaUlA4MTmUW7FjFz3czoz4mKeK5ZTOHxiY/zhzP/gKr8iMPME4xhfONk6NDaAhZVc1UjEZix9U9MnjG5xQvDt3O0Y1LWGXxV8B9CkSAjxnhZ/T9n1AVuNut5t1LCtq16lRh//dlDDZJqT0UVx268QEDPJ25orXfF4iRWLkvi57cUktGh7msx6VwPUgoWfl7d9awounfBaXFyqOLQSeF6NjBINPN3zefl/70crZ2cqJlvMFg9Z7+TqxO5XeqUEzdoIobxjZNzz4XcXD2sXwiJzR6JlueSEkJakIW750WrurQ0lw28DE+4hCX75jNyXAWeUpUtG6rn/FituuSj1wubN8OBA/G/HStCqaZ4deiwnmtZn5CHzyv46wsZZPUOcOEV9c9cO3YJMXCIlwWfJldzj1utukvbpJgIRAJUhFrek2BgcKIxKWsS5cFyVu5fCSQu4jkS0ScZUkoeXPBgtH2DxGEY3zhRVZg3D2bNgqtvOcLdj+6NlucKBmGTbwllwbJmVzCKlbFdx9LR2ZkvD/yb4bkVKIqMRj3XxOHQK8McOQKbNun/1gwUi4VgEIoKG5awnDmtHUcOmbnz/oJGFXfOnOJhzy4r2zYdU+kwm/XZdTCoG+Aib1H8HTUwOMkZ23UsdpM9GvWcKOMrhL5stfrAauZsmlMr5dCg+RjGtwmoKpx/PlxxQyEjxnqiIhGBAKw48jHt7O0Y23Vsq/RFEQq/GX0P53a6Fpdbo1+On1Ur6reKQuhycQ6HXqptyxY9sjger+6RI3o79c16d2238MGsVM65sJQBgxv3c592pgezRWP+p7WDfgMBcJqdHPIeMiTtDAxqYDPZGNdtHAt3LURKicmkr9NqzbhVNE1f9rFYYM6mOTjNzhbL2vgxYxjfBOILl7Ns/yKm9J7SKi7nSi7oP4Vxnc4gFIJR48rZvtlG4ZGGAyNUVZ8Fm0y6UP2OHbGlJgWDeqCVs56awZoGrzybicsV4YZf1Q6yqoskl8bo8RUsmeeqtl5lMuk5xKqiEtEirRLEZmBwojEpaxJW1cphr36/NTfoqrKYQnmwnM+3f855vc/Dbm4guMOgSRjGN0FEIpBsT2L2JbO5bvB1rX7+oOUg/9r6V4aP1fWQ1zQw+62K2ayvB4dCelT0nj0Ni3MUFekzXqWeWe8XHyezcZ2dG+88QnJK7K/fZ04po7TEFE3ZAn3dt/SovLNZNVPoa6Zsl4HBSciFfS9k3jXzopXTnM7mCeyEQvqL+WfbP8MX9vGzAUZub0tgGN8EEQjo0YH9MvrR1d218QMSzG7/Bmbl/ZEj7i9JSw/z+UfJvPNGGt8sdca0rmuz6f0vK9ODsvLza0dNSqmLatQ36y0tUXjzlQwGDvFy5vnxiTYNH1OBOzVczfWsqvqDIBjUBTcOVxw2XM8GBjVQFRUhBBFNv9EbykCIhcr1Xptq4/QepzM4c3ACemlQE0NkNkEcqijgzT0vcqfjl2Sltn5h7smnTCBlaQaf7pkN3MTG7+1sWmfHapP0y/ZFg8IaQgj9rVmTul7z4cN63dSUFH0NKBwGqdQ/633zlQy85Qp33n8oqgdbkyu/HEdxsHr5xVRLOrMmL+f0sz18MseNp0zBlaxF++TzgduioEmN8mA5yVZDEMrAoCpL8pZw35f3MefyOaRbujS5HSn1j9UKP+n7kxaTxzUwZr4JQUpYceRTPtnxIZLjk49qs5g5q+vFrD68hDItHxBIKfD7lGoKXLGgCN3tZLfrKllbt+ru31BIT7Gqiw3f2/jiIzcXXVVMj1Pq93nVNLxVv5t0XhmhkMJX84/l/JpM1V3PRypqH29g8GOnR0oPSgIlLNy1sFkaz8Ggfo9vLdrcYElRg+ZjGN8EEArpUc4DMwZWk3trbS4ZcAkSjXD29GrfV1XgiofKoCwh9KAsoM4ZbTisB1llZIa4+qb612Ubcxmf0jdAt6xANblJq1V3hUupu56P+I5E3WsGBgY63VO6c0raKSzYtQBF0e+bpgRdBYPgSApz00c3MXXB1MR31CCKYXwTwM7iHWwv28AFfS9oVjvNVXHqk9mNLDER1VXdAKoqdO8VaHK7Fovueq7Pbf2fWank7bDyy3sOYXfU/TuEtRAvfd/wzSwEnHleGRvX2TmwT9etVBQ9mC0Q0NOqNKnhCXoabMfA4GRBShnzc2FS1iRW7V9Fqb8Up7PpSlffFS/lsPcw5/U+r2kNGMSEYXwTwNKC/6IIhSm9pzS5jWAkSFgLN0uS0maDZ8b+leyCp7HZNYSQqKokHBb864127M1rpPxQEzicb2LmtHbknlrOmAn1pwLtq9jF0vx5jbZ3+jkehJAsqJHz6ztaW8FmsnG4IrYUJgODExVNapT5yyjxlxCRsXl6JmVNIiIjLNm9BKezaTNfKeGTHXNoZ2/HhO4T4m/AIGYM49tMNA2c5iTO731+NNS/KVSEKjAppmq1c+NFCEhvp/Dw8/v51ZPf8/NbC3nk+QM89PsDHMo386truvPJHHdCa37+9cUMpAa/vPdQnaIbIU1f/+3h6sMbEz4j1ZJea59kc2r0/zMywwwZ6WXBJ65oPy2WY+u+dpOdYn9xi+tmGxgcDyJahGJfMWWBMjKTMslun42gngjHGuRk5nDb8Nvo264vlvhXmQiFwCcKWbx7ERf2vRCzmviXdYNjGNHOzcQfgOsG3UiXpgcYIqVEILColmM/1ycf1QhuN/zkk2H4IxXQATia8eO+N52BH+fxyu8zWbncya9/V0BKWvPWTr9Z5mTFYhe/uOMwHTrVNobFgSP8buVNTOl2Bed1v4IMe0dmTV4e3V7oL+CGRWcxLGNcteMmnVfGC492ZMP3NrKH+LEcXffVlXcEUkrK/GWkOVquXKOBQWsSioQoD5ajCpVu7m6kO9Kjxk8IQTASjD4f6kMRCr8e82ugaXm+oRD8UL6QsBbm4v4Xx9+AQVwYM99msr98N0mu5hmxilAF6fZ0FKGQ7khvVhEBux3d8NagNHyEp/68n1t/fYhvv3Fw25XdWVmPBnQs+P2Cvzzfnm5ZAS6+unbhhHzvPn674ir2VeSRae9Ua7vHA8lqJhf1vJ7FB/7LttL10W3jJpZjtWnRwCtFHCsxCEddz17D9Wxw4hOMBCnyFRHV12chAAAgAElEQVSMBMlKzWJox6F0dHWsNutUhUogHFvMRkSLsGr/KnZ7tiOEnjYYK+EwXDbwEj664iN6t+sd769iECeG8W0GES3MM5uu5blVjzSrnVAkRPsk3WWdmZQZ841WFw0JqysKXHRVCX9+ew8paREeubszr/2hPQF//LPsd99Ko+CAmV/df6hWTd88z1Z+u+JKykIlPJv7D0a0H19te3m5Ho1ZUQGX9ryJZHMKb21+Mbrd7pCcekY5X813EQzofRNCr8oEuvEt8ZcQiiSgcKmBwXHAF/JR5CsiIiP0SevD4A6Dae9sX2e9XFVRY15mCWthbvnvLcz8YYYedBXnLWK3C/qm943vIIMmYRjfZrC28Gs84WIm9Ty9yW0EI0FsZhtOsz4LdZqdOCyOaH3OliDrlCB/nr6Hi64q5uPZKdx5bTe2b4l9kWjPLgvvz0jjzPNKGTTcV21babCIe7/+ORLJC2NmMiB1aLXtfr8uadmzpx5BLUIurup9O7vKtlDoL4juN2lKGRXlKv9bql+Xquu+QggEgrJAfCpaBgbHm4pgBUXeIsyKmf7p/RnUfhBpjjQUUf+juKFtNbGarJza9VQW7lqIza7FHHSlafCvvOf5w/+eivlcBs3jpDK+QohzhBBbhBDbhRAP1LPPZUKIjUKIDUKIfzXnfIsPfEyS2cX47uMb37kevEEvnZM6R9d4hRB0cXVpkSICERnhg13T8Ye9WKySW399mGde3Ue5R+Xu67sx+5+pjVZDkRJefa49NofGTXfVFrxwW9L4Rb/f8NLYWfRw9am2LRTSP1lZugHu1El3dZ3b9UreOv0L2tkyo/sOHuElvX0o6nq2WPQZc6VUps1s41DFoeZdkBOI1h7bBolDSkl5sJwibxFOs5PszGwGtB+A2+aOKbZDIDApppjz2yf1nMShikPs8a2PuWSox+fny4PvURooje0Ag2Zz0hhfIYQKvAacCwwArhRCDKixT29gKjBOSjkQuLsp5+rwQgfE44JFBz6mPOQh5685jHtzXOMH1kCTGghIsadU+z7FllJNqzVe2tlrRxSnWtJZX7iKv298llu+Oo9vChYBMCzXy99m5ZF7WjlvvpLBA7d34XB+/XF4Cz5zsW6Ng1/ccaRawNZXBz9jfdFqAKZ0u5wOjur61pqmu4179tRdzqD/27EjBLwWHKYkIlqY/RV5gJ5TfMY5HlZ97aSkSI1GUldd9y0LlDXLRX+i0Jpj2yBxaFKj1F9Kib8Et9XNoA6D6JPehyRLPTJxDZBsTSYQiW2sT+g+AVWoLDuwoN6ynzVZfvBLKsIeI9CqFTlpjC8wCtgupdwppQwC7wIX1tjnZuA1KWUxgJSySVOngoqCWt8d8cUve1gRrKC9s32t8oOqotLJ1YnyYHlTuseKG5fz6blb+GjyFj4/T//MmrycwemjeWHMO9hUB4+uvo2n1tzFEX8BySkav3vuIL9+OJ+tG2388qruLPmi9gOi3GNi2h8z6Jft49yfHntD/mT3LJ799tfM3vFGnf2RUo9W7tq1tjxlerpuhINBeG7tPUz95gaCRx8yZ0wpQ4sIFs/T5SaFqFH2UPBjcT232tg2aD7V0oWcmQzpMIReab1wmB1NbjPFlkIwHNtSVKo9leEdh7OmYGXMaYWLCubQKakzuZ1zm9xHg/g4mVKNOgN7q/y8D6g5kvoACCGWAyrwmJTy85oNCSFuAW4ByMzMZPHixTF1YNfaXXF1OKyFKTWVkifyACgvL4+eSyLxhrwUKUVxtVmJFoZwEGSN+I3etOOlvi/zYf4c3jvwLw6W7eTFAX9GCMHE0dD3zw5eeiGHZx/qxNcL9nPrLzdhtUb4dk0G774zjLISlUceXUnoiAcpJbMPvss7+99mpDuX33b5PwKHal+DSAScZijbq39qYjqqYDXJdRpfHfyM/2x4lQs7XExHF/TqlcqXH1o5d9IuzBIKd0LFAf04KSWrWY3dZK927doizexfq45ticQX8lGs1I5ibyoBbyDu+6O1SFTfNKmhSQ2BwKyaMSkmdh79rzmUl5fzw8of8IVj/5vc2flOUs2phLy7CPigoVThgkA+60v+x9Vdr2H397vj6lvYH2bFshUx5yIbHONkMr6xYAJ6AxOBLsBXQogcKWVJ1Z2klK8DrwOMGDFCTpw4sXorS+puPGtI7NWMKt2lOZk50e8WL15M1XNtPqKLmzfljdnngy1bwOygVjSyFbi6w1RO73015aFSbCk98Ye97KvI45QhWbw0/RCz3oww661ObNqUQXJKhN07rQQDAtUEM94ZzJN/3stbW3/PB/vf5ozOF/CbQc9gUmon5Xs8emHuHj1o0AW2dy8MLMliWNEnzM7/N1P630ySOZnJFwb420vtOejpS49eQcrKoPdAveACQLGvmMEdBvO/Zf+j1t+pDVHzb9sCJGZsowcBrj24ttZySHPYtXZXXPdHa9KcvoW1MOWBciSSZGsyHZI6kGxNrjNquaksXryY08afxuoDq0m1pzZ+AJCF/vvk5enPAput/n3VMhvndbuKGyb+gs7JnePq2/ZvtzP21LFYTU1Q9fiRczK5nfcDVRcauxz9rir7gI+klCEp5S5gK/oDq9XxhX10ctXOf61Kx6SOTa4sYrfDKafoa6SBepaKOjm70SdFN/6zd77BXct+xt83PkuICn5+ayEvvL6XQECwbZONYEABBJGwOFolyUGR/zA/7XEt9wx+rk7D6/XqgVJduzZseAEyM3X39PV97sETKom6sCee7UFRJQuryE36q1wSgaDUf9IHiZxQY7uliUfvuKXOXxlAFQgH6OruyuAOg+mf0Z9Ue2pCDW8lqqLiNDvjyoL45/f/5E8bpjYa8ZwsOnL/6EfiNrwGzeNkMr6rgN5CiCwhhAW4Avioxj7/QZ8ZIIRIR3fVxe0TynRm1vouvY4gp/qoDKRKsTU8s0i2JmNVrU3OZ01Kgt699WCnikaCp3+adR3ndLuUD3ZN55Yl5/F1/nwGDPZzzoU11lRNPvymfHZtc3DfkOe5dcCDdaZCBIP6eXv2PDZLbQiLBbp0gUxlAKd3Op+1hV+jSY2UtAgjx1Sw8HMXkYgeiFVeZSncYXGQX5Efw9U4oWm1sd3W8Ya8FPuKKfGX6P/6SvCFfK1ijIORIMW+YkoDpbitbga2H8iQDkPo6OqIzdTA1DJBpNhT4noZL/QWMi/vQzzBknr3yfNsZVPJ6noLohi0HCeN21lKGRZC/AqYh77m9ZaUcoMQ4glgtZTyo6PbzhJCbAQiwL1Syvpr4NVD/j36w37z4c2EtFDcLpeKUAUdnB0afUMWQtDJ1YndpbtJUZvmAqycAe/efcwFXBcus5u7cp7gzM4/5c/rH+XxNXdwTe876Z9zPza7xO8TYC2Fq36CcBbSPeu/qErdwycS0V1dvXs3LPpRk9RUOHIEbur9GClOR9SoT5pSxjfLOvH9agc5w7yUlECHDvoxFtVCsa/4uM6EWprWHNttlcpCAzazjUEdBmFVrfjDfipCFVGDWDkGrCYrVtWakBloRItQEaogrIVxmp30Su2F2+Y+LrrHLouL/VpNh0f9TOo5ib+t+RvfFi2mU7uf1rnPe9tfZ+WhJZw3ainQ8i8QBsc4aYwvgJTyU+DTGt89UuX/JfCbo5/jRjgSJsOZEdO+afY08krymqX3bLHoM9B9+6CoSNd/rq+pAWnDePXUuczd9Q9Gt59E514VhO/uANZjco4SeCUwinEsr3W8djSyOSsLnHGqVwqhz363bHGhAOWhMvwRH7mnCZxJERZ8msywXC9lZfrMutKwCyEIy5O70MKJMrZbgkA4QEWwgi7uLnRydYq+lDktTpwWJ+2d7ZFS4gv78Aa9lARKKPWXRlWhzKoZm8lWK6ugIbwhL4FwAEUodEjqQDtHu2ZFKycCu9ke1/7Z7bNp72zPqsKFnHfKT2vd8+WhMpbnf8GkTj8jqaFFYYMW4aQyvicC/rCfZGtyzDeSWTWT4cigJFDSpPzASlQVunXT03oOHtRnwPXV5zUpZi7rdUv057C1to5ySbDu1KqyUl08IzW2uJBaOBz6+m/B4Qi/XnMx3V29eWzEXxl/podF85L51f0CkPj9x4yv0+zkSOQIe0vrCKVuI2iyEfUSg1pIKSn1l2JWzQxsPxCXtR63DfoLmMPswGF2kO5MR0pJIBLAF/JR4i+hxF+CJ6JXDDMpJqwma61CBZXFDUBfEspKycJldcWlMNWSWFQLZtVMRIvENKtXhMIZWWfwn00f4Q0EcNqqe+iWHPiEoBbgglOM3N7jgWF8WxlvyEu/9H5xHZOZlKmrOcXhwq0LIXR3rcUCe/bohq5mJHRz8HggLU03ns2hfXsoKlI5s9MlzNj+MuuLVjPpvFP57D8pLF+cxNiJHsrKIPloDJZZNSOEaLPFFrwhr2F84yQYCeIJeOiQ1IGu7q5xzVpBN8Y2kw2byRaNEA5GgvhCPkr9pRQH9PVihO5aLvIVYVNt9EjpQYotpc1G77qtbsoCZTgtsbmVzj3lXAqKPZT4ynDaqnvb5u2dQzdHH4Z3zW6Jrho0gmF8W5GIFsGkmHBZ6n+Dr4tK91ogHEjIQyEtDSxW2LlDX59NhMfJ69XXl7t0aTyyuTFMJr2dM/3X8ened3hz0wu8OGY4HToFWfBJMqefrRvfqihCaZZnoCWJaBH8NC1q/cdIWaAMJPRP75/QdCeLasGiWnDb3HSjG6FICF/Yx0p1Jdnts3GanU1e2mkt3FY3R7xHcBKb8R3dZTR9naPZs6f69yWBIvK9e7mo6y+x29v273yy0jb8KT8SPEH9Tb4pgSCdXJ3whrwJ60uSE/r00Q1ledOEtKJUpjJ1716/Kzte3G5on2bnsh53sankO74+NJ9JUzysXeWg6IiJUKhpNUsN2i5hLUyht5BkazKDOgxKqOGtC7NqJtmajEkxkWRJavOGF/To/nixWOCgL6+a9yXFmsaMM5ZydufLE+r9Mogdw/i2IprUSHfEnpJUlRRbCopQmqz3XBc2G/Tqpbufa84kq5JqqVsrGvTCCMGgHmAVT2RzYwgBnTvD+IyL6JrUi+8Lv2HSlDKkFCz6XPcc+HyNNGJwwlAeLKc8WE7vtN70TuvdaOH4Hys2kw0hRFzR/V/t+5L/W302W0t+APQCK5rUkGEL6Sn2ZnuqDJqG4XZuJXwhHym2lCbnA6qKSkdXRw56DuK2uRPWL7NZV586cEBP80l268XrqzJrsh7VHDi0C2v7Y0pAmtRnzT176gY80Vit0LmjiSe1d+mQmgyEGDDIx/xPkjn/0mJKS/UZssGJS0SLUBooJdmSTM+Mnq2SL3siowgFl8VFMBKMeQlqTPeRKKisKFhAv9TB/K9gIdM2/p4HB/6DUR26tXCPDerDmPm2Er6Qj45JHZvVRrojvUUCd1RVX2Pt1AlKS4ipDJmUemRz584tawDT0yHFnkwwCAe9exl/3kH27LKyL89KaSkxC8cbtD0qg5+6u7vTP6O/YXhjJMUWn9hGii2F7HYj+Dp/AQBf7J1DUAuQYesUrTBm0PoYxrcVCGthLKqlwVSJWLCZbKTaUhO69luJEHqUclaWPpttTJLO49ENY0Zs6cpNRlV1eco9xQe4efG5lPR5BbNZY+FnyWha/dKZBm0XKSXFvmKEEAzqMIiOro4nxHprW8FpcSKJ761zQtdJ7K3Yzvqi1aw6/BVndr4Ik2JKSLClQdMwjG8rUB4op6OrY0LyBTu4OjRZ7zkWUlN1ZSq/v/411fJyXbqyc+fmRzbHQlIS9OnYiWFpE/ho/zSGnbGbRfNchMPGuu+JRiAcoMhXRCdXJwZmDDzuwhUnInaTPW5Ft0lZkwB4+tv/Q5MRTu9wMU4nKIYFOG4Yl76FkVIikaTZ0xLSnsviapbecyw4ndC3r35j1oyEDgT0deJu3Vr3xu3YEa7K+g3+sA85/ilKi01sXOukpH7ZWoO2hIRSfymhSIjs9tl0dXdtkQIEPwbMqhmrao0qeMXCjZ9fDkBxQBfH+eWKc7hm4bgW6Z9BbBjGt4Xxhryk2lITlrQvhKBzcmfKQ83MD2oEq1WPhHY6dRezlPonGNQDtFo7PcFigVG9e3F65iV8G56Bq9tWlnyRjMejF3AwaHtEtAiBcIDyYDlhLUw7RzsGdRjU7OUXA3Db3PEVWfDVVqQrCtStUmfQOhjGt4UJRAJ0cHVIaJuptlQE8aUbNIXKSOjUNCgpOVal6HitE6Wlwc/7/gqraqPXOfP531In5R6lWolBg9YjokUIRoJUBCso9ZdS4tMrDVV+/GE/JsVEqi0Vm9lGz9SecStVGdRNii2lRb1fBi2PcSe0IKFICKtqjVvRqjEq9Z6LfEUtPotQFOjSGWxWKNmtr78eL4SAwadk8mpwEWVd0vm/1xX+t8RFn74nfT3f44ImNcJamFAkRFgLo0kNgQChL6eYFBN2s50UWwoOsyOqPWxWzJhVc7UYh91i93H8TU4+bCab/rcwOGExjG8LUh4sp0dKjxaJ5GzvbE9BRUHC260LIfSo5vLYq5m1GA4HdO/o4og5QObQVXw1fwAXXl5quHASiEBgUk1UBCuixtVutmNVrfUaV4PWxWaygdBfkIy/w4mJYXxbiEqXcKWoe6JxWpwkWZISpvd8IpGZCZ9vnU/BhXdQMGMe27f2pndiPfs/asyqmSEdhhgP9TaMEAK31U0gEogpPzrdns6RGuu+6famqe0ZJAbD+LYQFaEK2jnatahMXqekTmwr2vajM74mE/xk0Hje3NKFwjPv56v58+h11fHu1cmFYXjbPim2FHaX7o7J+C6/cTnBIGzerP/cty+GwMZxxrjDWohQJESms5m19RrBbXOjKmpC9Z5PFDLSLNww8G7ouJZPt37KOzN6smieMyZ1roaIRGDRPCevvZCWsPaWznczc0YW//1vbOphBgax4DA74gq6NJv1jAUhEqvDbtA0jJlvCxCMBLGZbC1e4k5VVDoldWK/Z39C9Z5PBISAy4eex/PfPYD37F/wL+Bf24Bt0M6WzsJrlsfdZiQCo/52GmHrIbDpbSWsvR7w9hpgDWQ6M8m/Jz/u9gwMqmI32+PaXwi97KfF0jriOAYNI1o6XSUW1qxZ095kMr0BZNPGZuOFhYXdO3asW5M5FAkhkbWiDjWpYVJMcYsI+P1+bHHm8UgpCUaCreImDIfCmMxt533N7xcUheqOAjNFjpWjUzQrirQg0YioFbX2VTQbijQT0TRCtrrbswW7IokQUWtLe6qaHSFNSBEmohyT3JIaaOZSUqx6X0qDpVFt7u4p3QHYv39/MCMj42CMv3JCaWhsJ5qmjO3Woi33DRruXzCi19WMNfI5EtENb6IEcsLBMO4UN926dsNcI/lfCLFGSjkiMWc6+WgTT1KTyfRGhw4d+mdkZBQrinL83waqsHHjxu79+/evc5sv5EMiqxs+qZfsspvtcRtEj8eDyxV/6pA/5NejHltYcirgDWB1tJ2FokP5KorSuLizUt4RxZuJVP1E2m2ovb2sE4o/g4hagWxXd3umQwPQzB601K21jy/pjhJ0o1lK0FJ2VNumlranfWp7kt3JeINe9lbsBaB/J31MRSKRcHZ29nFRO2hobCeapo7t1qAt9w0a7l8gHCCshWN+0a+cayVq5uuv8BMKhti3bx9ZWVmNH2AQpU0YXyC7LRrephCREUyKqVUDVsyqGV/Yh9K2nAYtjt0uoR7b28N5zKiIJIE4uqMmaxubyu3eCjP59YzAbj0DgLme4xUEASQ2ZJXtXq9CUWQL7lQ3QghsatudXRmcmKiKSkiLXWwj0e5mIQRp7dI4csRQy4qXtvK0Vk4Gwwsgkc1S8SkoKODtt9+O6xhFKCgoLa541dZISq5fV9JmVaMfq0XBYgGLRVT7vuZ2d0r916/h4wUWC1hrbE9N0Z90RsUenaaMbYOGUYRyXMU2JNIY302krRjfhFBUVJS8bt267HXr1mXv27evVuZnQUFBu++++27w+vXrB6xfv35Afn5+NNFNVdXh/fr1G9C7d++B5557bk+PxxP3tZFSd0FXuoAeeeQR5s+fX+/+f/vb3/jnP/8Z/dnj8XD33XdzxhlnxHVeIQRm1UxExh5KO+PtGRw4cCCu8yQKKSURLVLtZeGJx55g4YKFcbUjBJgUE1vWb2H5gmMBUSbFhMlE3B+zmTpfnJrTXl0PRrMSvzD28R7bNWmtsd0Upk+fftzGdl00dq0aYu3atcybN6/e7ZXj63i8eFcqnhlpaU2jTQRcff/993mDBw+O2W8RDsPs2bi//RbHsGF4L72UUlWV/PDDD9m9e/fearVaQxs3buyflZW10+l0RpV/CwoK2nm9XmdWVtaemm06HI6hXq/3O4ALLrgga9iwYd7HHnusYOPGjcMHDBhAOBzGZKr+YK655hvWwlEVoKbQnLUnKSXekFd/E47hTfSsSWfx7HPPMnzE8FrbIpEIqlp7Dak5a75SSjSpRQPUVEUlrIX1WXszb94Zb89g1Ter+PNf/tysdlqCnVt30rdfX5AQDAax26tHqK5fv96bnZ29qfLntjK2E01rrqtOnDiRF154gREjasf61DW22/Ka7/Tp01mxYgWvv/56vfv4w/7jonQV1sJofo3k5GQ2bdpEzfgBI+CqYU64V5ZwGE47jd433kjPF1+k04030vO00+hdXOxxWiyWgN1uDyqKIlNSUoqKi4tTGm+xNqeeemr59u3brf/9739dP//5z7ngggsYMGAAkUiEe++9l5EjRzJo0CDemPZG9JgX/vACucNyGTFsBA888AAA119/Pe+//z4ADzzwAAMGDGDQoEHcc889ADz22GO88MILgP6Ge8YZZzBo0CAuuugiiouLAf1Bcv/99zNq1Cj69OnD0qVL6+zzCy+8wISxExg5bCRPPv4kALvzdjMkZwi333o7wwYP4/xzz8fn8zF3zly+XfMtv7juF+QOz8Xn89H3lL48NPUhxowcw5z35zD/y/lMOHUCY0aO4aorrqL8aG3Bvqf05cnHn2TMyDGMGDKCLZu3ALBq5SomnDqB0SNGM/G0iWzdshUpJW9Pf5tLLr6E8845jwF9BvDW39/ib6/8jTEjxzBp/CQKCwuJaBFuvuFm5s6ZC8C3a75l8hmTGTtqLD+Z8hMOHtSDgc+adBYPTX2IU8ecSs6AHJYtW0YwGOSJx59g7ty55A7PZfa/Z1NUVMSlP7uUkUNHMn7ceH5Y90NThkGr05bG9t///vfoMc899xw5OTkMHjz4uIzt559/PtqvRx99FIC8vDz69+/PzTffzMCBAznrrLPw+Xy8//77rF69mquvvpohQ4bg8/no0aMH999/P8OGDWP27Nl88cUXjBkzhmHDhnHppZdGx3aPHj149NFHGTZsGDk5OWw+qkixcuVKxowZw9ChQxk7dixbtuhjfvr06fz0pz9l8uTJ9OjRg1dffZWXXnqJoUOHMnr0aIqKimpdqzVr1jBhwgSGDx/O2WefHR3bdV2LYDDII488wty5cxkyZAjvvfceRUVF/PSnP2XQoEGMHj2adevWoQq11We+Uuov0YbLuem0lYCrKDfcQNf166m3wnZxMerOndgry8j5fCgrV5Kcm+uwud3dhaLQF0DKTJOUEVVRcGdn433rLfYClJaWpvzwww9JVqvV361bt702m61atEIoFOKzzz7LGDt2rFpQUODauHEj06dPp0ePHrz22mvYbDYWLlxIIBBg8uTJjD91PNu2bePjDz9m/vz5JDmTKCoqwuPxEAqF8Pl85OXlMWfOHNasWYMQgpKSEjweD4FAALPZjMfj4ZprruG5555j/PjxPPXUUzz00EM899xzRCIRvF4vCxYsYN68eTzyyCN89NFH1a7JggUL2LhxIwsWLCAiI1x1xVUs/HIhXbp0Yfu27UybNo2XX36Z66+7ntnvzubyyy9nyNAhPPXkUwwdNhSkfjO5XW4WL1lMYWEh11xzDR988AFOp5M/vvxHXnr+Je679z59v2R9vzemvcGLf3iRV159hR7devDpp59iMplYvGgxv5v6O2bMnIEW0ti4fiPLli0jEAgwZMgQHn/8cb766iseeOAB3pn+Dr+8/ZdEIhHCgTDlpeX8+q5f869Z/yI9PZ25c+by8IMP89prr6FpGgF/gAULFvDFF1/w1GNP8eFHHzJ16lS++/Y7XnhRf9jfe++9ZA/MZuaMmSxZsoQbrr+BZcuWJXKYxoyUUn8wSrjnHhPr1lWvwerzZdkqx2xbGttnnXUWY8eOZevWrcydO5f58+fjcDiO69iWUnL55Zfz+eef06VLF7Zt28Ybb7zBSy+9xHXXXcfMmTO54oorGDp0KE899RTDhg0jHA4jpSQpKYklS5ZQWFjI1VdfHR3bL7/8Mq+88gpTp06ttt+0adN49tlnefXVV+ncuXN0bC9atIj77ruPmTNn4vf7WbduXb1j+/XXX+eOO+6IXquioiJuv/123n33XdLT05kzZw733Xcff/nLX+q9Fg8++CBr1qzhpZdeAuCee+5hwIABzJgxgyVLlnDNNdewbNkyNDTCxF7fN1FITeLxePD7/SxevLjVz38i0+aMb2N4vag167dqGni9QnG7afD1Ly0trSQjI6NIURSZn5+fvmvXrqz+/ftvBQgEAkq/fv0GAOTm5noefvjhffPnz0/Kycnpk5OTA8BXX33FunXr+PjjjwEoKS1h977dLF22lGuuv4b27dujKmrUhWU2m7Hb7XTp0gWHw8Hdd9/N+eefz/nnn4/FYsFqtWK1WtE0jbKyMsaPH4/L5eKWW27h0ksvxeVyoaoqV1xxBS6Xi9NOO40HHniglots2bJlLFq0iAkTJqBJjYqKCvbs3UOv3r3okdWDkaNHAjB81HAOHDyA1WFFURTMNnPUjSyE4PKrL8fqsLJ20Vq2btnKueecC0AwFCQ3Nw3Ui08AABrESURBVBeh6G+6P7vsZ1gdVkaOHsknn36CxW7Be9jL7Xfczo7tO1AUhXAoTLIrGbvdzqRJk+jUqRMAbrc7+rsNHz6cdevWkexKxqSaUCwKeXvy2LRpExdfdDGguwk7dOwQ7fPFl1yM1WFl1JhRPPDAA1gdVswWM0KI6O+y8puVzPr3LKwOK2edexa33347gXCA5OTkuMdbcxHi2OxACFDVmrfcsXDttjS2S0tLOXjwICtWrOCmm24iM1NXazteY3v8+PEAlJeXs3//fvr160dWVhbjxukF4XNzcykoKIi263Q6o20JIbj22mtxuVwsWbKELVu2cM455wD6UsCIESNwuVwIIbjqqqtwuVyMGzeOTz/9FJfLRUlJCTfccAPbtm1DCEEoFMLlcmGz2Rod2y6XK3qtDhw4wKZNm7jooosAfWx37NixwWths9kQQkR/l5UrVzJnzhxcLhfnn38+v/zlL/UKU3ZTq5ZrDGth7CY73gpvtJ9Dhw5ttfOfDLQ541v5Fl8fs2bhvvFGevp8x1zmdjvaM88E8nNz9yT369dvG8C+fQUdALp06RKVEjKbzdGIpMzMzCMHDhzoUvmz1WrVNm/evLHm+RyOY5NwKSWvvPIKZ599NnBszXf+l/OrBVrVxGQysXLlShYsWMD777/Pq6++ysKFsQcXWY+KsKqqSjhc++1WSsnUqVO59dZbCWvhaB3V3Xm7o8dWHu/31V/81ul0Rts748wz+OfMf1bbHvAGov3RpAYKhMK60MjTTzzNmWecyccffkxeXh4TJ06MGp2qfVAUJfqzoiiEw2GEEPrLgKIHjfUf0J8ly5Y06Vq0ZZ5/PoTdXv2WW79+lz87O3sLtK2xXUlDwT7QumO7Knl5ebXGts/nq3l4lKpje/LkycyaNSu6zePxNNifhx9+mNNPP50PPvggOrZr7g91j+2av8vAgQP5+uuv6+xjU8e2ECLqem4NN7CUEoXmx2r82Dnhrt6ll1I6eDDldjvaUbk0bfBgyq+80lYQCARsPp/PommaKCkpSUtNTS2pemwgEIhGQhUWFqZYrda4yrCfffbZ/PWvfyUU0r1527Zuo6KigtPPOJ2Zb8/E69XVjyrXeiopLy+ntLSUKVOm8PLLL/P9999X2+52u0lNTWXFihUAzJgxgwkTJsTVr7feeovy8nJUoXJw/0EKChouN+hyuaJrXTUZlTuKr1d8zY7tumBERUUF27Zsi24Pa2EEAotiQRUqDrMDT5mHLl305/306dNj7ntVVEVl0IBBHDl8JHotQqEQGzfUshvVSHIl4Sk/9gAde+pY3p31LgBfLfmK9Hbpx2XWGy9taWxv3bqViooKJk+ezD/+8Y82MbYB9u/fz6FDhxo8xuVyVTOoVRk9ejTLly9n+/btwNGxvW1bnftWUlpaSufOnYGmj22Avn37cvjw4ajxDYVCbNhQW/SlKjXv09NOO4133nkHgMWLF5Oero9tVVGj6mktTURGMKtmY723mbS5mW9jmEywdCnbZs/G/d13OIYO1SNCTSaFrl277tm2bVsfgLS0tCNOp9O/Z8+eTk6ns6Jdu3al+fn57cvKylKEEFJV1XBWVlZePOe+6aabyMvLY9iwYUgpaZfejnfff5ezzj6Lzes3M2LECCwWC1OmTOGZZ56JHufxeLjwwgvx+/1IKaPrN1V5++23ufnmm7n33nvp2bMn//jHP2Lu11lnncWmTZsYM2YMoL/lT/vHNCzm+tXTr7n2Gu68407sNjuLly2Ofi+lJD09nb9N+xvXXnMtgYA+2334sYfp1rUbAoHD7MButmOq4kK97777uO6663jqqac477zzYu57Tew2O7Pfn81dd91FaWkpkXCEX931KwYMHFDvMRMmTuD53z9P7vBc7rn/Hn73yO+49eZbGTl0JHaHnWlvTWtyf1qTtjS2MzIy+M9//sM555zD2rVr28zYTkpKYubMmXVG41dy/fXXc9ttt2G322vNMjMyMpg+fTpXXnlldGw/9NBDDBs2rN72EjW2LRYL77//fnRsh8Nh7r77bgYOHFjvMaeffjpPP/00Q4YMYerUqTz22GPccMMNDBo0CIfDEc2bVoSCbHhlIjEcPUW80rkGtTkhU41ak8p0jLrwhXx6epHJmpDSgYlKedCkhjfkxSRM1Jd/L6VEIqP/Vs1HrUz/qfwIIRAIysvLWy0lQ0pJSAsRjARRhdroW3Zbk76sJN5Uo9akobGdaNpyOk9b7hvE1r/oPd/C676a1FBQsJlt1fpmpBrFzwk3822LmETbuoyKUDApJiJaRFe+qsfIqkJFUZWoSk6lkW0L7iQhBBbVgoKCP+I31pgMDBqg8h5u6XVfTWpYTEY9wkTQtqzGCYhJMbV4QYOmYFbMuorUUREQRTlmZGMV4mgLmFQTdmEnEAkQ0SKGu8vAoB5URUXTtBa7tytze1Vh3IOJwDC+zUAI0arh/fGgKnog1IliZBtCVdSoAQ5r4Zjc0AYGPzZUoRIm3GIFVjSpGYFWCaTtTdlOICqjfdsqJ9NNIoTAqupr6zV1oQ0MDGjxZRmJbHNLbCcyxpVsBm3R3XwyE10HFgqBcMAQdTcwqEK0wpGk3kDLpqJJTY8RMZ55CcO4kgYnHCbFhN2kRw1HtNgrORkYnMwIob+MaiQ+31eTWpOqcRnUj2F82xhGzdPYUBQFu8kerY5k0PYxxnbLoyqJL7JQ2Z4R7JhYDON7lB9bzdO6WL16NXfddVeTj29uHdW8vDyys7Nj3r/qOrBErxGcqM+iRYu46IKLmt1OqwgfNIIxtk+8sd1UVKE2Oua+WvIVF194ccxtVs56T6YYkraAseZ7lKr6txdccEHWiy++mPHYY49FNRqbUvP0iSeeaHD7bbfdVu1nl8tVTXO2tRkxYkSdNVBjZfr06WRnZ0eF5luDynVgv/BjM9nq3Ke++sQNUbm2XF+bMffv6Lp0a0n/1YUxtk/MsR0LNcd2SxhIiWyzWR0nMifkzDf9D+mDxeNieNVP+h/SByeq/ROx5mld51+7di2jR4+Ouc3Fixdz/vnn1+obQHZ2Nrt3746rjuqCBQsYOnQoOTk53HDDDVE5v6qsWbOGwYMHM3jwYF577bXo9w1d60ry8vLo168fV199NSNHjOTyyy4n4A+gKiq9evbiwakPMnLESObOmcukMybx3bffoSoqxUXF9OrZS3ejSXjg/gcYnTuaoUOG8sa0N1AVFVVR8Xg8XPCTCxjQfwB33H6HnuOoqPzqjl+ROyqXQTmDeOLxJ6L7P/TgQ+Rk5zB0yFDuv+9+VEWtVtUoFoyxXZuWHtu5ubnk5eW12bE9YsQILrnkkqi+ds36xBMnTmT16tUAFBUWMbDPQKSURCIRpt4/lXGjxzFy6EjeeP1Y/fGysjIuuuAiBg0cxJ2334l2tJzWXXfcxbjccQwbPIwnH38yWkThwakP1vobGzSPNml8R00b1bfm5/fLfp8B4Al4lEJfYa3XsMrvDnoOmmoeG8+5Q6EQ8+bNS87JyfEBbNq0iT/96U9s3bqVN998E7fbzapVq1i1ahXTpk1j165dfPbZZ3z44Yd88803fP/999x3333V+1ZYyAcffMCGDRtYt24dv/vd72qd99prr+WJJ55g3bp15OTk8Pjjj0e3hcNhVq5cyR//+Mdq31dS3/mvvfZannvuuSa12RDbtm3jjjvuYMOGDaSkpDBnzhwuueQSRowYwTvvvMPatWsRQnD99dfz3nvv8cMPPxAOh/nrX/9aq61f/OIXvPLKK7UE+eu71jXZsmULt99+O6tXryY5OZm//OUv0W3t2rXj22+/5Yorrqj3d2noPCtXruSVV15h48aN7Nixg7lz5wL/397ZB1dV3nn88014CSImAbc71NAS14rFWIghvCijUKuzjhZXQaxbXRBdR0cryy7bbdedLq7rjN2xLNvWvrBqEF+qVCzL6rqitYDgKkGUGEBGhYBSUUAjRRshyW//OM8NN+GGJDfJvSfJ7zNzJ+flefmec785z3mec+7zg7vuuouNGzdSVVXFmjVrqKqqatd3DDBlyZRmn9lrZue5t93bXe3tRmtkyQNLyM/PZ/3L61n38joq7q+gZmcNABsrN7Jw0UJeq3qNHTt2sOI3KwBYcOcC1r+ynspNlby49kU2V23mYO1BVqxY0aa3nY4Ry8Y3GyRinp511lmji4qKDs+dO3c/RD2+4uJiAFatWsXSpUsZO3YsEyZM4MCBA7z11ls8//zzXHfddU0h2oYOHdqs7Pz8fPLy8rj++ut58sknm4VygyhqSm1tLZMnTwZg1qxZrF27tmn/FVdEz2fKysqoqak5Rnuq+hNlJiLIdLTM41FcXMzYsWOPm3/79u0UFxdz+umnp6wfoLa2ltra2qZYrddee23TvtbOdUtGjBjRFNM1EVg8wVVXXdXmsRyvnvHjx3PqqaeSm5vL1Vdf3VT2smXLOPvssyktLWXLli1s3bq1ze84m7i3209v8LZQFOr0+ed55OFHmFA2gfPOOY8DHx1oiuY0rnwcxacWk5uby8yrZvLS+ijq1PJfL2dS+SQmjpvItq3beHPbmwwtHBpbb/dkYjmQv+GvN2xvbd+QgUOO+/Bs+JDh9cfL3xo9OeZpOrRVZr9+/ZqGogDq6uqOyZvIf7w4qunS2rluScsh3eT1RAxXaH48ycfSWj2rV69OWfbOnTu55557qKyspLCwkNmzZ1NXV9fu73j17NXN1qurq+tKSkr2gXvbvd2crvD2wkULufCiC5uVs3bN2pRl1+ysYdG/L2Ld/62jsLCQG+bcwJHPjzCg/4BOfcdOarzn2wHiGvM0Vf2JMhPPvDpa5siRI9m0aRMAmzZtSjks1pLkOKqjRo2ipqam6U47Vf0FBQUUFBQ03dEn4pRC6+e6Jbt3724KG/foo4829bBSHc+rr74K0PSssq16NmzYwM6dO2lsbOTxxx9n8uTJHDx4kMGDB5Ofn88HH3zAM888A7T9HcedvuztXbt2tZmnp3lbiAsvvJDFv1x8TPxxiIada3bW0NjYyBO/foJzzj0n8vYJR7296tlV5Cinx3s7rsSy59sWwwYNq2/5bGzYoGHd/mPPuMY8ba3+Bx98kJtuuonPPvus3WUm7oinT5/O0qVLOfPMM5kwYULTENvxaBlHtaKigiuvvJL6+nrKy8uPeQMWoKKigjlz5iCJiy66qGl7a+e6JaNGjeLee++lsrKSkpISbr755pTa5s+fz8yZM1m8eHGzmKzHq6e8vJxbb72Vt99+m6lTp3L55ZeTk5NDaWkpZ5xxRrNhwfZ8x+3Bvd2cTHj7tNNOazNvT/N2jnKYNWcWu3btYlL5JIwoTvey5csAKBtXxry583jnnXc4//zzuewvLiMnJ4cxY8cwpmQMRUVFTJw0kRzldJm3neZ4PN826EsxT5cvX87KlStbnQgh2/paUlNTw6WXXkp1dXXstCVIxDltaGjweL4x9Xa2taWis94+0hBiYac5MUZDYwP9c/u3Gafc4/mmT4/s+Tpdz8qVK7n99tt54IEHsi3FcbqUvujtzs557kEUuh8/uw4A06ZNY9q0admW0SFGjhxJdXV1tmU4MacvejtHOWnPruZBFDJDXM5uY2Njo89d5vRK4vBox+lbJIIspOO9jgZRcH+nR1wa3+p9+/blewPs9Dby8vLYv38/9fX13pNwMkqucjs8rWlHgyiYGQcOHCAvr3PTsPZFYjHsXF9ff8PevXvv27t3bwnxuSEAohl8MjWheF1dXaxNHGd9cdVmZhw6dIg9e/YwYMCAY7y0d+/efg0NDSdnQ5t7OyLO2iB9fY3WyJGGIx16/msYOcpp91zOdXV1FBQUUFRU1GF9fZ1YNL5lZWUfArF8KDNu3DhLzJva3axevZrS0tKM1JUOcdYXZ20Q6ZsyZcox20ePHv1Gtt4IdW9HxFkbpK+vrr6OzXs3UziosN15PvrjR5R8oYQTB5zYrdqcmPUyO4ukP5e0XdLbkr53nHTTJZkkfw3e6RG4t52OMjB3ILk57R96PtxwmLzcPAb3H9x2YqfT9JrGV1IucC9wMTAauFrSMT9ilDQEmAu8klmFjpMe7m0nHSSRPzCfz+uPjbiUik8Pf8rwIcM9bm+G6DWNLzAeeNvMdpjZYeAx4LIU6e4EfgjUpdjnOHHEve2kRUFeAXX1bdsh8aJVR4aonc4Ri2e+XcQpwLtJ6+8BE5ITSDobGGFmT0v6+9YKknQjcGNYPSSpw5PZp8nJQCxn+grEWV+ctUHr+r7cjrzu7e4lztqgM/pEDv3Jo5GGNtMZjdTTvm5y+7S1x9t9lt7U+B4XSTnAQmB2W2nNbDGwuLs1tUTSxjhPxxZnfXHWBt2rz73dOeKsDeKtL87a4k5vGnbeA4xIWi8K2xIMAUqA1ZJqgInASn8xxekBuLcdp5fRmxrfSuArkoolDQC+BaxM7DSzT8zsZDMbaWYjgZeBaWaWmd9aOE76uLcdp5fRaxpfM6sHbgWeBbYBy8xsi6R/kRTL3xCnIOPDgR0kzvrirA06oc+93e3EWRvEW1+ctcWaWIQUdBzHcZy+RK/p+TqO4zhOT8EbX8dxHMfJMN74xgBJIyT9TtJWSVskzc22ppZIypX0mqSnsq2lJZIKJD0h6U1J2yRNyramBJLmhe+0WtKvJMV3Bv8uxn3dOeLsa+jb3u4KvPGNB/XA35nZaKKfidySavrALDOX6GWfOPIfwP+a2RnAGGKiU9IpwG3AODMrAXKJ3lTuK7ivO0csfQ3u7a7AG98YYGbvm9mmsPwHon+yU7Kr6iiSioBLgPuyraUlkvKB84D7AczssJnVZldVM/oBgyT1A04Afp9lPRnDfZ0+PcDX0Ie93RV44xszJI0ESonX5PiLgO8CHYvMnRmKgX1ARRg+vE9SLMKymNke4B5gN/A+8ImZrcququzgvu4wsfU1uLe7Am98Y4SkE4HlwN+Y2cFs6wGQdCnwoZm9mm0trdAPOBv4uZmVAp8CrYbcyySSCokCIBQDXwQGS7omu6oyj/s6LWLra3BvdwXe+MYESf2JLlCPmNmT2daTxLnAtDBt4WPA1yU9nF1JzXgPeM/MEj2qJ4guWnHgG8BOM9tnZkeAJ4Fzsqwpo7iv0ybOvgb3dqfxxjcGKAqgeT+wzcwWZltPMmb2fTMrCtMWfgt4wcxic4drZnuBdyWNCpsuALZmUVIyu4GJkk4I3/EFxOilme7GfZ0+Mfc19HFvdwV9JqpRzDkXuBZ4Q9LrYds/mtn/ZFFTT+I7wCNh3uMdwHVZ1gOAmb0i6QlgE9Gbv6/Rt6bjc193jlj6GtzbXYFPL+k4juM4GcaHnR3HcRwnw3jj6ziO4zgZxhtfx3Ecx8kw3vg6juM4TobxxjfLSLolTELgOL0K97bjtI43vt2EJJP0o6T1+ZIWtEhzDTDMzA5lWl9rSKqRdHK2dWQaSQskzQ/LsyV9Mdua4op7u2fh3o4n3vh2H58DV7Txz54L3NkdlYfJzp30mE00ZZ6TGvd2z2U27u1Y4I1v91FP9KPzeS13SFoiaYaZPWhmJulQ2D5F0hpJ/yVph6S7JX1b0gZJb0j6s5DuTyQtl1QZPueG7QskPSRpPfCQpJGSXpBUJem3kr6UQsswSatCXM77ACXtuybU/bqkX0rKTZG/XNJLkjaHtENCvS9K2hQ+5yQd31pJT0vaLukXknLCvp9L2hh03JFU/t2K4sFWSbonRf1DJa0I+1+W9LWw/URJFeG8VUmaHrYfSso7Q9KSFuXNAMYRTW7wuqRBkn4QznO1pMVhRh8k3Zak7bFWfNAbcW+7t53OYmb+6YYPcAg4CagB8oH5wIKwbwkwIzlt+DsFqAWGAwOBPcAdYd9cYFFYfhSYHJa/RDR9H8AC4FVgUFj/b2BWWJ4DrEih88fAD8LyJYABJwNfDfn7h30/A/6qRd7EzDvlYf0kolnTTgDywravABuTjq8OOJWoZ/Rc4jwAQ8PfXGA18DVgGLCdo5PBFKTQ/xPgn8Py14HXw/IPE+crrBcmn+uwPANYknTu5ofl1URxSknWFpYfAr4Zln8PDGxNW2/9uLfd2/7p/MeHb7oRMzsoaSlR0Ok/tjNbpZm9DyDpHSARpusNYGpY/gYwOtykApykoy+2rDSzRF2TgCvC8kPAv6Wo77xEGjN7WtLHYfsFQBlQGeoZBHzYIu8o4H0zq0wcb9A9GPippLFAA3B6Up4NZrYjpPsVMJlo0viZkm4kusANB0YTzWVbB9wv6SngqRT6JwPTQ/0vhN7OSeEcNQX3NrOPU+RtL1MlfZfowjsU2EJ08a4i6kWsAFZ0ovweh3vbve10Dm98u59FRPOfViRtqycM+YehqQFJ+z5PWm5MWm/k6PeVA0w0s7rkisKF5NMu0i3gQTP7fhp55wEfAGOItCbrbDmfqUkqJuo9lZvZx2G4LM/M6iWNJ7pYzgBuJeoBdIbk+vPaSiwpj6hnNM7M3lX0YlEi3yVEF/hvArdLOsvM6jupryfh3nZvO2niz3y7GTP7CFgGXJ+0uYbozhtgGtC/g8WuIpp0HYBwF56Klzh6h/xt4MUUadYCfxnKuRgoDNt/C8yQ9IWwb6ikL7fIux0YLqk8pBmi6GWYfKJeQyPRxPrJz9PGSyoOF+argHVEQ3qfAp9I+lPg4lDeiUC+RRPxzyO64LXkxXBsSJoC7A+9lOeAW5LOUeK4PpD01VD/5SnKA/gDMCQsJy5G+4OeGaG8HGCEmf0O+IdwzH3qZzXubfe2kz7e+GaGHxE9a0rwn8D5kjYTDZ919I7+NmBceBliK3BTK+m+A1wnqYroQjE3RZo7gPMkbSEaotsNYGZbgX8CVoX8zxENmTVhZoeJLjI/CcfyHNE/9M+AWWHbGS2OrxL4KVH4sZ3Ab8xsM1FUlDeJnvmtD2mHAE+F+tcBf5tC/wKgLKS5G5gVtv8rUBheJNnM0WHN7xEN8b0EvJ/6tLEE+IWiSDyfE31f1cCzQT9EF92HJb0RtP/YzGpbKa83494+invbaTce1cjJGOHufb6ZXZptLY7Tlbi3nY7iPV/HcRzHyTDe83Ucx3GcDOM9X8dxHMfJMN74Oo7jOE6G8cbXcRzHcTKMN76O4ziOk2G88XUcx3GcDOONr+M4juNkmP8HjR57qk2YSfoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eAnmbB_pO18c"
      },
      "source": [
        "# Perceptron Multicapa (Capítulo 12)\n",
        "\n",
        "En esta sección se construye el perceptron multicapa estudiado en el capítulo 12, la construcción del perceptron se realiza en la clase *NeuralNetMLP* "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "collapsed": true,
        "id": "If8-JMKDX2jg"
      },
      "source": [
        "import numpy as np\n",
        "import sys\n",
        "\n",
        "\n",
        "class NeuralNetMLP(object):\n",
        "    \"\"\" Feedforward neural network / Multi-layer perceptron classifier.\n",
        "\n",
        "    Parameters\n",
        "    ------------\n",
        "    n_hidden : int (default: 30)\n",
        "        Number of hidden units.\n",
        "    l2 : float (default: 0.)\n",
        "        Lambda value for L2-regularization.\n",
        "        No regularization if l2=0. (default)\n",
        "    epochs : int (default: 100)\n",
        "        Number of passes over the training set.\n",
        "    eta : float (default: 0.001)\n",
        "        Learning rate.\n",
        "    shuffle : bool (default: True)\n",
        "        Shuffles training data every epoch if True to prevent circles.\n",
        "    minibatche_size : int (default: 1)\n",
        "        Number of training samples per minibatch.\n",
        "    seed : int (default: None)\n",
        "        Random seed for initalizing weights and shuffling.\n",
        "\n",
        "    Attributes\n",
        "    -----------\n",
        "    eval_ : dict\n",
        "      Dictionary collecting the cost, training accuracy,\n",
        "      and validation accuracy for each epoch during training.\n",
        "\n",
        "    \"\"\"\n",
        "    def __init__(self, n_hidden=30,\n",
        "                 l2=0., epochs=100, eta=0.001,\n",
        "                 shuffle=True, minibatch_size=1, seed=None):\n",
        "\n",
        "        self.random = np.random.RandomState(seed)\n",
        "        self.n_hidden = n_hidden\n",
        "        self.l2 = l2\n",
        "        self.epochs = epochs\n",
        "        self.eta = eta\n",
        "        self.shuffle = shuffle\n",
        "        self.minibatch_size = minibatch_size\n",
        "\n",
        "    def _onehot(self, y, n_classes):\n",
        "        \"\"\"Encode labels into one-hot representation\n",
        "\n",
        "        Parameters\n",
        "        ------------\n",
        "        y : array, shape = [n_samples]\n",
        "            Target values.\n",
        "\n",
        "        Returns\n",
        "        -----------\n",
        "        onehot : array, shape = (n_samples, n_labels)\n",
        "\n",
        "        \"\"\"\n",
        "        onehot = np.zeros((n_classes, y.shape[0]))\n",
        "        for idx, val in enumerate(y.astype(int)):\n",
        "            onehot[val, idx] = 1.\n",
        "        return onehot.T\n",
        "\n",
        "    def _sigmoid(self, z):\n",
        "        \"\"\"Compute logistic function (sigmoid)\"\"\"\n",
        "        return 1. / (1. + np.exp(-np.clip(z, -250, 250)))\n",
        "\n",
        "    def _forward(self, X):\n",
        "        \"\"\"Compute forward propagation step\"\"\"\n",
        "\n",
        "        # step 1: net input of hidden layer\n",
        "        # [n_samples, n_features] dot [n_features, n_hidden]\n",
        "        # -> [n_samples, n_hidden]\n",
        "        z_h = np.dot(X, self.w_h) + self.b_h\n",
        "\n",
        "        # step 2: activation of hidden layer\n",
        "        a_h = self._sigmoid(z_h)\n",
        "\n",
        "        # step 3: net input of output layer\n",
        "        # [n_samples, n_hidden] dot [n_hidden, n_classlabels]\n",
        "        # -> [n_samples, n_classlabels]\n",
        "\n",
        "        z_out = np.dot(a_h, self.w_out) + self.b_out\n",
        "\n",
        "        # step 4: activation output layer\n",
        "        a_out = self._sigmoid(z_out)\n",
        "\n",
        "        return z_h, a_h, z_out, a_out\n",
        "\n",
        "    def _compute_cost(self, y_enc, output):\n",
        "        \"\"\"Compute cost function.\n",
        "\n",
        "        Parameters\n",
        "        ----------\n",
        "        y_enc : array, shape = (n_samples, n_labels)\n",
        "            one-hot encoded class labels.\n",
        "        output : array, shape = [n_samples, n_output_units]\n",
        "            Activation of the output layer (forward propagation)\n",
        "\n",
        "        Returns\n",
        "        ---------\n",
        "        cost : float\n",
        "            Regularized cost\n",
        "\n",
        "        \"\"\"\n",
        "        L2_term = (self.l2 *\n",
        "                   (np.sum(self.w_h ** 2.) +\n",
        "                    np.sum(self.w_out ** 2.)))\n",
        "\n",
        "        term1 = -y_enc * (np.log(output))\n",
        "        term2 = (1. - y_enc) * np.log(1. - output)\n",
        "        cost = np.sum(term1 - term2) + L2_term\n",
        "        return cost\n",
        "\n",
        "    def predict(self, X):\n",
        "        \"\"\"Predict class labels\n",
        "\n",
        "        Parameters\n",
        "        -----------\n",
        "        X : array, shape = [n_samples, n_features]\n",
        "            Input layer with original features.\n",
        "\n",
        "        Returns:\n",
        "        ----------\n",
        "        y_pred : array, shape = [n_samples]\n",
        "            Predicted class labels.\n",
        "\n",
        "        \"\"\"\n",
        "        z_h, a_h, z_out, a_out = self._forward(X)\n",
        "        y_pred = np.argmax(z_out, axis=1)\n",
        "        return y_pred\n",
        "\n",
        "    def fit(self, X_train, y_train, X_valid, y_valid):\n",
        "        \"\"\" Learn weights from training data.\n",
        "\n",
        "        Parameters\n",
        "        -----------\n",
        "        X_train : array, shape = [n_samples, n_features]\n",
        "            Input layer with original features.\n",
        "        y_train : array, shape = [n_samples]\n",
        "            Target class labels.\n",
        "        X_valid : array, shape = [n_samples, n_features]\n",
        "            Sample features for validation during training\n",
        "        y_valid : array, shape = [n_samples]\n",
        "            Sample labels for validation during training\n",
        "\n",
        "        Returns:\n",
        "        ----------\n",
        "        self\n",
        "\n",
        "        \"\"\"\n",
        "        n_output = np.unique(y_train).shape[0]  # number of class labels\n",
        "        n_features = X_train.shape[1]\n",
        "\n",
        "        ########################\n",
        "        # Weight initialization\n",
        "        ########################\n",
        "\n",
        "        # weights for input -> hidden\n",
        "        self.b_h = np.zeros(self.n_hidden)\n",
        "        self.w_h = self.random.normal(loc=0.0, scale=0.1,\n",
        "                                      size=(n_features, self.n_hidden))\n",
        "\n",
        "        # weights for hidden -> output\n",
        "        self.b_out = np.zeros(n_output)\n",
        "        self.w_out = self.random.normal(loc=0.0, scale=0.1,\n",
        "                                        size=(self.n_hidden, n_output))\n",
        "\n",
        "        epoch_strlen = len(str(self.epochs))  # for progress formatting\n",
        "        self.eval_ = {'cost': [], 'train_acc': [], 'valid_acc': []}\n",
        "\n",
        "        y_train_enc = self._onehot(y_train, n_output)\n",
        "\n",
        "        # iterate over training epochs\n",
        "        for i in range(self.epochs):\n",
        "\n",
        "            # iterate over minibatches\n",
        "            indices = np.arange(X_train.shape[0])\n",
        "\n",
        "            if self.shuffle:\n",
        "                self.random.shuffle(indices)\n",
        "\n",
        "            for start_idx in range(0, indices.shape[0] - self.minibatch_size +\n",
        "                                   1, self.minibatch_size):\n",
        "                batch_idx = indices[start_idx:start_idx + self.minibatch_size]\n",
        "\n",
        "                # forward propagation\n",
        "                z_h, a_h, z_out, a_out = self._forward(X_train[batch_idx])\n",
        "\n",
        "                ##################\n",
        "                # Backpropagation\n",
        "                ##################\n",
        "\n",
        "                # [n_samples, n_classlabels]\n",
        "                sigma_out = a_out - y_train_enc[batch_idx]\n",
        "\n",
        "                # [n_samples, n_hidden]\n",
        "                sigmoid_derivative_h = a_h * (1. - a_h)\n",
        "\n",
        "                # [n_samples, n_classlabels] dot [n_classlabels, n_hidden]\n",
        "                # -> [n_samples, n_hidden]\n",
        "                sigma_h = (np.dot(sigma_out, self.w_out.T) *\n",
        "                           sigmoid_derivative_h)\n",
        "\n",
        "                # [n_features, n_samples] dot [n_samples, n_hidden]\n",
        "                # -> [n_features, n_hidden]\n",
        "                grad_w_h = np.dot(X_train[batch_idx].T, sigma_h)\n",
        "                grad_b_h = np.sum(sigma_h, axis=0)\n",
        "\n",
        "                # [n_hidden, n_samples] dot [n_samples, n_classlabels]\n",
        "                # -> [n_hidden, n_classlabels]\n",
        "                grad_w_out = np.dot(a_h.T, sigma_out)\n",
        "                grad_b_out = np.sum(sigma_out, axis=0)\n",
        "\n",
        "                # Regularization and weight updates\n",
        "                delta_w_h = (grad_w_h + self.l2*self.w_h)\n",
        "                delta_b_h = grad_b_h # bias is not regularized\n",
        "                self.w_h -= self.eta * delta_w_h\n",
        "                self.b_h -= self.eta * delta_b_h\n",
        "\n",
        "                delta_w_out = (grad_w_out + self.l2*self.w_out)\n",
        "                delta_b_out = grad_b_out  # bias is not regularized\n",
        "                self.w_out -= self.eta * delta_w_out\n",
        "                self.b_out -= self.eta * delta_b_out\n",
        "\n",
        "            #############\n",
        "            # Evaluation\n",
        "            #############\n",
        "\n",
        "            # Evaluation after each epoch during training\n",
        "            z_h, a_h, z_out, a_out = self._forward(X_train)\n",
        "            \n",
        "            cost = self._compute_cost(y_enc=y_train_enc,\n",
        "                                      output=a_out)\n",
        "\n",
        "            y_train_pred = self.predict(X_train)\n",
        "            y_valid_pred = self.predict(X_valid)\n",
        "\n",
        "            train_acc = ((np.sum(y_train == y_train_pred)).astype(np.float) /\n",
        "                         X_train.shape[0])\n",
        "            valid_acc = ((np.sum(y_valid == y_valid_pred)).astype(np.float) /\n",
        "                         X_valid.shape[0])\n",
        "\n",
        "            sys.stderr.write('\\r%0*d/%d | Cost: %.2f '\n",
        "                             '| Train/Valid Acc.: %.2f%%/%.2f%% ' %\n",
        "                             (epoch_strlen, i+1, self.epochs, cost,\n",
        "                              train_acc*100, valid_acc*100))\n",
        "            sys.stderr.flush()\n",
        "\n",
        "            self.eval_['cost'].append(cost)\n",
        "            self.eval_['train_acc'].append(train_acc)\n",
        "            self.eval_['valid_acc'].append(valid_acc)\n",
        "\n",
        "        return self"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FYj_QBe8LK7D"
      },
      "source": [
        "### Calidad del Vino Rojo\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TS8H_kGtNCVO"
      },
      "source": [
        "from numpy import genfromtxt\n",
        "data_with_header = genfromtxt('/content/drive/MyDrive/Dataset/winequality-red.csv', delimiter=';')\n",
        "data = data_with_header[1::,:]\n",
        "x = data[:,0:-1]\n",
        "y = []\n",
        "for label in data[:,-1]:  y.append(int(label)) \n",
        "\n",
        "y_binary = []\n",
        "for label in data[:,-1]:  \n",
        "  if label >= 6: y_binary.append(0)\n",
        "  else: y_binary.append(1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fGAJjEnMNiqd"
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "X_train, X_test, y_train, y_test = train_test_split(x, y_binary, \n",
        "                                                    test_size=0.3, \n",
        "                                                    random_state=1, \n",
        "                                                    stratify=y_binary)\n",
        "sc = StandardScaler()\n",
        "X_train_std = sc.fit_transform(X_train)\n",
        "X_test_std = sc.fit_transform(X_test)\n",
        "\n",
        "X_train = np.array(X_train_std)\n",
        "X_test = np.array(X_test)\n",
        "y_train = np.array(y_train)\n",
        "y_test = np.array(y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "865fUMAHRZkl"
      },
      "source": [
        "#Parámetros Perceptron\n",
        "neuronas_ocultas = 15;\n",
        "epocas = 150;\n",
        "constante_aprendizaje = 0.00001;\n",
        "\n",
        "Perceptron_vino = NeuralNetMLP(n_hidden=neuronas_ocultas, \n",
        "                  epochs=epocas, \n",
        "                  eta=constante_aprendizaje,\n",
        "                  shuffle=True,\n",
        "                  seed=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nzYFZUzjL3kO",
        "outputId": "ec65e066-248b-4fa2-b44d-e7a83a2fcf7f"
      },
      "source": [
        "Perceptron_vino.fit(X_train=X_train, \n",
        "                    y_train=y_train,\n",
        "                    X_valid=X_test,\n",
        "                    y_valid=y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "150/150 | Cost: 1545.25 | Train/Valid Acc.: 53.44%/53.54% "
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<__main__.NeuralNetMLP at 0x7f21333ee5d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 135
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "zoBrFBF1Rtv9",
        "outputId": "39ae356c-a5d8-46fa-e893-30de26d3ed31"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(range(Perceptron_vino.epochs), Perceptron_vino.eval_['cost'])\n",
        "plt.ylabel('Costo')\n",
        "plt.xlabel('Epocas')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "id": "X4nVYCENR7RO",
        "outputId": "2596ba14-75ea-4643-c005-a31e3fa45d18"
      },
      "source": [
        "plt.plot(range(Perceptron_vino.epochs), Perceptron_vino.eval_['train_acc'], '--')\n",
        "plt.plot(range(Perceptron_vino.epochs), Perceptron_vino.eval_['valid_acc'], '--')\n",
        "plt.ylabel('Precisión')\n",
        "plt.xlabel('Epocas')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "T0hQuowWTNkB"
      },
      "source": [
        "## Clasificación de vidrios"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d9U158GyTbsP"
      },
      "source": [
        "import scipy.io as sio\n",
        "data = sio.loadmat('/content/drive/MyDrive/Dataset/data_glass.mat')\n",
        "data = data['data']\n",
        "x = data[0:146,0:-1]\n",
        "y = []\n",
        "for label in data[0:146,-1]:  \n",
        "  if int(label) == 2:\n",
        "    y.append(0) \n",
        "  else:\n",
        "    y.append(1)\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(x, y, test_size=0.2, random_state=1, stratify=y)\n",
        "sc = StandardScaler()\n",
        "X_train = sc.fit_transform(X_train)\n",
        "X_test = sc.fit_transform(X_test)\n",
        "\n",
        "X_train = np.array(X_train)\n",
        "X_test = np.array(X_test)\n",
        "y_train = np.array(y_train)\n",
        "y_test = np.array(y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yrCsy0JPT7Hw"
      },
      "source": [
        "#Parámetros Perceptron\n",
        "neuronas_ocultas = 25;\n",
        "epocas = 100;\n",
        "constante_aprendizaje = 0.001;\n",
        "\n",
        "Perceptron_vidrios = NeuralNetMLP(n_hidden=neuronas_ocultas, \n",
        "                                  epochs=epocas, \n",
        "                                  l2=0.01,\n",
        "                                  eta=constante_aprendizaje,\n",
        "                                  shuffle=True,\n",
        "                                  seed=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Gd96DqfvUc2t",
        "outputId": "ee0cfc06-ce29-4d32-b7bb-5e896ef2df3b"
      },
      "source": [
        "Perceptron_vidrios.fit(X_train=X_train, \n",
        "                        y_train=y_train,\n",
        "                        X_valid=X_test,\n",
        "                        y_valid=y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100/100 | Cost: 154.60 | Train/Valid Acc.: 62.93%/66.67% "
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<__main__.NeuralNetMLP at 0x7f2126088ed0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 158
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "jK-_XP-dU0Q2",
        "outputId": "1abf5732-8bb8-40b8-ce13-43d042ae5913"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(range(Perceptron_vidrios.epochs), Perceptron_vidrios.eval_['cost'])\n",
        "plt.ylabel('Costo')\n",
        "plt.xlabel('Epocas')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "NbjgDXKpU1O0",
        "outputId": "1a5a2b50-6cfc-47d2-9d7e-bf0a99687ad5"
      },
      "source": [
        "plt.plot(range(Perceptron_vidrios.epochs), Perceptron_vidrios.eval_['train_acc'], '--')\n",
        "plt.plot(range(Perceptron_vidrios.epochs), Perceptron_vidrios.eval_['valid_acc'], '--')\n",
        "plt.ylabel('Precisión')\n",
        "plt.xlabel('Epocas')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0GuQjU1eHpho"
      },
      "source": [
        "## Parametro de regularización"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OpeNwB9HIqM4"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rUp6MlbQI3nj"
      },
      "source": [
        "from numpy import genfromtxt\n",
        "data_with_header = genfromtxt('/content/drive/MyDrive/Dataset/winequality-red.csv', delimiter=';')\n",
        "data = data_with_header[1::,:]\n",
        "x = data[:,0:-1]\n",
        "y = []\n",
        "for label in data[:,-1]:  y.append(int(label)) \n",
        "\n",
        "y_binary = []\n",
        "for label in data[:,-1]:  \n",
        "  if label >= 6: y_binary.append(0)\n",
        "  else: y_binary.append(1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G_UtGfVnI5jh"
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "X_train, X_test, y_train, y_test = train_test_split(x, y_binary, \n",
        "                                                    test_size=0.3, \n",
        "                                                    random_state=1, \n",
        "                                                    stratify=y_binary)\n",
        "sc = StandardScaler()\n",
        "X_train_std = sc.fit_transform(X_train)\n",
        "X_test_std = sc.fit_transform(X_test)\n",
        "\n",
        "X_train = np.array(X_train_std)\n",
        "X_test = np.array(X_test)\n",
        "y_train = np.array(y_train)\n",
        "y_test = np.array(y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AAjN4HE2JI_Y",
        "outputId": "8fc6bddc-9cd1-416b-a7e5-7545049ad424"
      },
      "source": [
        "neuronas_ocultas = 15;\n",
        "epocas = 100;\n",
        "constante_aprendizaje = 0.00001;\n",
        "L2 = np.round(np.arange(1.5,2.5,0.1),2); #tercera\n",
        "#L2= np.round(np.arange(0.0,4.5,0.5),2); #busqueda media\n",
        "#L2= np.round(np.arange(0.0,10,1),2); #busqueda inicial \n",
        "acc_train = np.zeros(len(L2))\n",
        "acc_test = np.zeros(len(L2))\n",
        "cost = np.zeros(len(L2))\n",
        "k = 0;\n",
        "for i in L2:\n",
        "  Perceptron_vino = NeuralNetMLP(n_hidden=neuronas_ocultas, \n",
        "                      l2=i,\n",
        "                      epochs=epocas, \n",
        "                      eta=constante_aprendizaje,\n",
        "                      shuffle=True,\n",
        "                      seed=1)\n",
        "  Perceptron_vino.fit(X_train=X_train, \n",
        "                      y_train=y_train,\n",
        "                      X_valid=X_test,\n",
        "                      y_valid=y_test)\n",
        "  acc_train[k]=Perceptron_vino.eval_['train_acc'][-1]\n",
        "  acc_test[k] =Perceptron_vino.eval_['valid_acc'][-1]\n",
        "  cost[k] = Perceptron_vino.eval_['cost'][-1]\n",
        "  k +=1"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100/100 | Cost: 1547.50 | Train/Valid Acc.: 53.44%/53.54% "
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 554
        },
        "id": "HPTzgfPaL95k",
        "outputId": "17bd8bde-485f-49cb-c6a7-c02a624beb05"
      },
      "source": [
        "plt.plot(L2, acc_train, '--', label='Entrenamiento')\n",
        "plt.plot(L2, acc_test, '--', label = 'Validación')\n",
        "plt.ylabel('Precisión')\n",
        "plt.xlabel('Parametro de regularización')\n",
        "plt.legend()\n",
        "plt.show()\n",
        "\n",
        "plt.plot(L2, cost, label = 'Costo')\n",
        "plt.ylabel('Costo')\n",
        "plt.xlabel('Parametro de regularización')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5Zq4qpBu3UOA"
      },
      "source": [
        "# Perceptron Multicapa (Capítulo 13)\n",
        "\n",
        "En esta sección se construyen redes neuronales profundas con ayuda de la API Keras de TensorFlow conceptos estudiados en el capítulo 13. Comenzamos importando las librerias de TensorFlow, "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EvDyKYlx4aOG"
      },
      "source": [
        "import tensorflow as tf\n",
        "import tensorflow.compat.v1 as tf\n",
        "tf.disable_v2_behavior() \n",
        "import keras as keras\n",
        "np.random.seed(123)\n",
        "tf.set_random_seed(123)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pETfcmSM5Jvu"
      },
      "source": [
        "El siguiente paso es cargar los datos, si los datos tiene categorias o etiquetas numeradas (1,2..) o (clase1, clase2, ..) es necesario llevar estas categorias a la representación one-hot, sistema en el cual la salida de la red neuronal sera un vector de $s$ posiciones donde $s$ es el número de clases. Las etiquetas tendran la forma de vectores con entradas cero en todas las posiciones menos en la posición $i$ que representa la clase a la cual pertenece. \n",
        "\n",
        "Por ejemplo para un sistema de tres clases identificadas con $(1,2,3)$, los vectores asociados en representación one_hot serán:\n",
        "\n",
        "$$(1,2,3)= [1 \\ 0 \\ 0]^T; [0 \\ 1 \\ 0]^T; [0 \\ 0 \\ 1]^T $$\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RvWatfxu4fUL",
        "outputId": "5b653291-f98e-4c15-b4a2-08fc3f3b20e3"
      },
      "source": [
        "from numpy import genfromtxt\n",
        "data_with_header = genfromtxt('/content/drive/MyDrive/Dataset/winequality-red.csv', delimiter=';')\n",
        "data = data_with_header[1::,:]\n",
        "x = data[:,0:-1]\n",
        "y = []\n",
        "for label in data[:,-1]:  y.append(int(label)) \n",
        "\n",
        "y_one_hot = keras.utils.to_categorical(y)\n",
        "\n",
        "y_binary = []\n",
        "for label in data[:,-1]:  \n",
        "  if label >= 6: y_binary.append(0)\n",
        "  else: y_binary.append(1)\n",
        "\n",
        "print(y_one_hot)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[0. 0. 0. ... 0. 0. 0.]\n",
            " [0. 0. 0. ... 0. 0. 0.]\n",
            " [0. 0. 0. ... 0. 0. 0.]\n",
            " ...\n",
            " [0. 0. 0. ... 1. 0. 0.]\n",
            " [0. 0. 0. ... 0. 0. 0.]\n",
            " [0. 0. 0. ... 1. 0. 0.]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NK3nmx4H6q9E",
        "outputId": "7e41a2ed-e3b1-443f-9164-7247921020e4"
      },
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "X_train, X_test, y_train, y_test = train_test_split(x, y_binary, \n",
        "                                                    test_size=0.3, \n",
        "                                                    random_state=1, \n",
        "                                                    stratify=y_binary)\n",
        "sc = StandardScaler()\n",
        "X_train_std = sc.fit_transform(X_train)\n",
        "X_test_std = sc.fit_transform(X_test)\n",
        "\n",
        "X_train = np.array(X_train_std)\n",
        "X_test = np.array(X_test)\n",
        "y_train = np.array(y_train)\n",
        "y_test = np.array(y_test)\n",
        "\n",
        "print(X_train.shape[1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "11\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PYesXoON7EY-"
      },
      "source": [
        "Coonstrucción del modelo en Keras"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gkLBWhV06db9"
      },
      "source": [
        "model = keras.models.Sequential()\n",
        "\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=22,    \n",
        "        input_dim=X_train.shape[1],\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='tanh'))\n",
        "\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=44,    \n",
        "        input_dim=22,\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='tanh'))\n",
        "\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=22,    \n",
        "        input_dim=44,\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='tanh'))\n",
        "\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=22,    \n",
        "        input_dim=22,\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='tanh'))\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=22,    \n",
        "        input_dim=22,\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='tanh'))\n",
        "\n",
        "model.add(\n",
        "    keras.layers.Dense(\n",
        "        units=1,    \n",
        "        input_dim=22,\n",
        "        kernel_initializer='glorot_uniform',\n",
        "        bias_initializer='zeros',\n",
        "        activation='sigmoid'))\n",
        "\n",
        "\n",
        "sgd_optimizer = keras.optimizers.SGD(\n",
        "        lr=0.001, decay=1e-7, momentum=.9)\n",
        "\n",
        "#model.compile(optimizer=sgd_optimizer, loss='categorical_crossentropy') \n",
        "#model.compile(optimizer=sgd_optimizer, loss='sparse_categorical_crossentropy')\n",
        "model.compile(optimizer=sgd_optimizer, loss='binary_crossentropy')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4tMjAbGY7J74",
        "outputId": "508456dc-7f15-4988-c23c-e6d7f270f5c7"
      },
      "source": [
        "history = model.fit(X_train, y_train,\n",
        "                    batch_size=5, epochs=10,\n",
        "                    verbose=1, \n",
        "                    validation_split=0.4)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 671 samples, validate on 448 samples\n",
            "Epoch 1/10\n",
            "671/671 [==============================] - 0s 334us/sample - loss: 0.0423 - val_loss: 0.8135\n",
            "Epoch 2/10\n",
            "671/671 [==============================] - 0s 343us/sample - loss: 0.0374 - val_loss: 0.8600\n",
            "Epoch 3/10\n",
            "671/671 [==============================] - 0s 364us/sample - loss: 0.0458 - val_loss: 0.8524\n",
            "Epoch 4/10\n",
            "671/671 [==============================] - 0s 354us/sample - loss: 0.0423 - val_loss: 0.8343\n",
            "Epoch 5/10\n",
            "671/671 [==============================] - 0s 351us/sample - loss: 0.0883 - val_loss: 0.8864\n",
            "Epoch 6/10\n",
            "671/671 [==============================] - 0s 356us/sample - loss: 0.0358 - val_loss: 0.8832\n",
            "Epoch 7/10\n",
            "671/671 [==============================] - 0s 362us/sample - loss: 0.0425 - val_loss: 0.8712\n",
            "Epoch 8/10\n",
            "671/671 [==============================] - 0s 350us/sample - loss: 0.0466 - val_loss: 0.8906\n",
            "Epoch 9/10\n",
            "671/671 [==============================] - 0s 364us/sample - loss: 0.0358 - val_loss: 0.8795\n",
            "Epoch 10/10\n",
            "671/671 [==============================] - 0s 341us/sample - loss: 0.0311 - val_loss: 0.8759\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VHXl0f9t7a0Z",
        "outputId": "15d817ec-28a9-480d-dc8b-b5425c1b3501"
      },
      "source": [
        "y_train_pred = model.predict_classes(X_train, verbose=0)\n",
        "y_train_pred_array = np.zeros(len(y_train_pred))\n",
        "k=0\n",
        "for x in y_train_pred:\n",
        "  #print(x)\n",
        "  y_train_pred_array[k] = x[0]\n",
        "  k+=1\n",
        "#print(y_train_pred_array)\n",
        "correct_preds = np.sum(y_train == y_train_pred_array, axis=0) \n",
        "train_acc = correct_preds / y_train.shape[0]\n",
        "\n",
        "\n",
        "print('Training accuracy: %.2f%%' % (train_acc * 100))\n",
        "\n",
        "y_test_pred = model.predict_classes(X_test, verbose=0)\n",
        "y_test_pred_array = np.zeros(len(y_test_pred))\n",
        "k=0\n",
        "for x in y_test_pred:\n",
        "  y_test_pred_array[k] = x[0]\n",
        "  k+=1\n",
        "correct_preds = np.sum(y_test == y_test_pred_array, axis=0) \n",
        "test_acc = correct_preds / y_test.shape[0]\n",
        "print('Test accuracy: %.2f%%' % (test_acc * 100))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training accuracy: 92.67%\n",
            "Test accuracy: 60.00%\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/tensorflow/python/keras/engine/sequential.py:450: UserWarning: `model.predict_classes()` is deprecated and will be removed after 2021-01-01. Please use instead:* `np.argmax(model.predict(x), axis=-1)`,   if your model does multi-class classification   (e.g. if it uses a `softmax` last-layer activation).* `(model.predict(x) > 0.5).astype(\"int32\")`,   if your model does binary classification   (e.g. if it uses a `sigmoid` last-layer activation).\n",
            "  warnings.warn('`model.predict_classes()` is deprecated and '\n"
          ],
          "name": "stderr"
        }
      ]
    }
  ]
}