numbers(edit 2).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f4db1cce",
   "metadata": {},
   "source": [
    "(Second Edit) Beginning the data collecting by loading in MNIST original, a dataset with 70,000 small images of digits. The images are 28x28 pixels, and are grayscale. The images are already split into a training set (60,000 images) and a test set (10,000 images). The labels are the digits these images represent: integers from 0 to 9."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "c74dda04",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.io import loadmat\n",
    "mnist = loadmat(\"mnist-original.mat\")\n",
    "mnist_data = mnist[\"data\"].T\n",
    "mnist_label = mnist[\"label\"][0]\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3698ad22",
   "metadata": {},
   "source": [
    "Once the data has been loaded in, preprocessing can begin"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "42424757",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "import numpy as np\n",
    "#preprocessing the data.  We must first reshape the data to a 4D array(samples,width,height,number of colors), where each image is a different 2D matrix\n",
    "#Normalizing the values of the data, could experiment with minmax scaler too.  To do this, we must flatten the data to a 2D array, then turn it back into a 4D array\n",
    "#I do not need to look for null values because this data does not have any\n",
    "mnist_data = mnist_data.reshape(-1, 28, 28, 1)\n",
    "mnist_data_2d = mnist_data.reshape(-1, 28*28)\n",
    "scaler = StandardScaler()\n",
    "mnist_data_normalized = scaler.fit_transform(mnist_data_2d)\n",
    "#After the 2D data has been normalized, return it to a 4D array\n",
    "mnist_data_normalized = mnist_data_normalized.reshape(-1, 28, 28, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b3e70cb3",
   "metadata": {},
   "source": [
    "The data must be split into a testing and a training set. The training set will be used to train the model, and the testing set will be used to test the model. The testing set will be 20% of the data, and the training set will be 80% of the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "de82273a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(56000, 28, 28, 1) (56000,)\n",
      "(14000, 28, 28, 1) (14000,)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "#Splitting the data into training and testing sets\n",
    "X_train, X_test, y_train, y_test = train_test_split(mnist_data_normalized, mnist_label, test_size=0.2, random_state=42)\n",
    "print(X_train.shape, y_train.shape)\n",
    "print(X_test.shape, y_test.shape)\n",
    "#print(X_train[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "05782353",
   "metadata": {},
   "source": [
    "We have 2D representations of the matrixes, they are 28 by 28 pixels. We can visualize the data by reshaping the data into a 28 by 28 matrix, and then plotting it using matplotlib. The data of these images are normalized."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "3ce2f77c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaAAAAGdCAYAAABU0qcqAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAAfUUlEQVR4nO3df2xV9f3H8VdbyuWH7WWltreVggXlt9SMQcdURGko3UYAyeKvJWAMRlbMkDkNi4JsS7rhN5vRMP1ng5kI/kgEItlYFGyJrrDxK4SNVWB1QGjLZKEXCv1Be75/EO92hYLnw+15394+H8lJ6L333fO+n5721dN77ps0z/M8AQAQsHTrBgAAfRMBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABP9rBv4sq6uLp06dUpZWVlKS0uzbgcA4JPneTp37pwKCwuVnt79eU7SBdCpU6dUVFRk3QYA4AadOHFCw4YN6/b+pAugrKwsSdLo0aOVkZFh3A0AwK/Ozk59+umnsZ/n3emxAFq7dq1eeuklNTY2qqSkRK+++qqmTp163bov/uyWkZFBAAFAL3a9l1F65CKEt99+W8uXL9eqVau0b98+lZSUqLy8XKdPn+6J3QEAeqEeCaBf/epXWrx4sR577DGNHz9er7/+ugYNGqTf/e53PbE7AEAvlPAAam9v1969e1VWVvbfnaSnq6ysTLW1tVc8vq2tTdFoNG4DAKS+hAfQ559/rs7OTuXn58fdnp+fr8bGxiseX1VVpXA4HNu4Ag4A+gbzN6KuWLFCzc3Nse3EiRPWLQEAApDwq+Byc3OVkZGhpqamuNubmpoUiUSueHwoFFIoFEp0GwCAJJfwM6D+/ftr8uTJ2r59e+y2rq4ubd++XdOmTUv07gAAvVSPvA9o+fLlWrhwob7xjW9o6tSpevnll9XS0qLHHnusJ3YHAOiFeiSAHnzwQf373//WypUr1djYqDvvvFPbtm274sIEAEDfleZ5nmfdxP+KRqMKh8MaN24ckxDgLMkO6z6B4cH4Qmdnpw4fPqzm5mZlZ2d3+zjzq+AAAH0TAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEz0yDRu9S5CDO4PaV1dXVyD7CVKQX6dkHiyanh7c780u65DMa5dsOAMCAJgggAAAJgggAIAJAggAYIIAAgCYIIAAACYIIACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJhgGnaKcZmYHOSUZZcp1UE+p6CmaH/3u9/1XXPLLbf4rpk/f77vGkkaPHiw75pIJOK7ZvXq1b5rNm7c6LvGdUJ1kJO3/UqFqdvJu7oAgJRGAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABMNIk1hQQzhdB3C67KuzszNp9yO5rcX999/vuyYvL893TWtrq++at99+23eNJIVCId814XDYd8348eN919x7772+a3bs2OG7RpIyMjJ817gMCQ1y6GkyDTHlDAgAYIIAAgCYIIAAACYIIACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJhpEGxGXIpUtNkIM7g9qXS82lS5d810jSmDFjfNcMHjzYd019fb3vGpfn5Drksr293anOr1tvvdV3TUFBge8al+GqktsAWJcBpi5SYYApZ0AAABMEEADARMID6MUXX1RaWlrcNnbs2ETvBgDQy/XIa0ATJkzQhx9++N+d9OOlJgBAvB5Jhn79+ikSifTEpwYApIgeeQ3oyJEjKiws1MiRI/Xoo4/q+PHj3T62ra1N0Wg0bgMApL6EB1BpaanWr1+vbdu26bXXXlN9fb3uuecenTt37qqPr6qqUjgcjm1FRUWJbgkAkIQSHkAVFRX63ve+p0mTJqm8vFx/+MMfdPbsWb3zzjtXffyKFSvU3Nwc206cOJHolgAASajHrw4YMmSIRo8eraNHj171/lAo5PwmMQBA79Xj7wM6f/68jh075vTuZQBA6kp4AD3zzDOqqanRZ599pj//+c+aP3++MjIy9PDDDyd6VwCAXizhf4I7efKkHn74YZ05c0Y333yz7r77bu3atUs333xzoncFAOjFEh5Ab731VqI/ZdJxGcLpMljUhcvgTtfeghoSGlSNJP3zn//0XTN06FDfNR0dHb5ruruS9Fpchp5Kl/907ldbW5vvmrlz5/qucblStqyszHeNJG3ZssWpLpkFNSz1q2AWHADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABM9/h/SJTOXoaI3UueXy5DQIAelBjWM1GVwZ3t7u+8ayW0I5+bNm33XuKxdkFwGVvbr5//Hyb59+3zX5OTk+K4ZMGCA7xrJ7XsjPd3/7/VB7Udy+xmRlpbmtK/r4QwIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYIIAAgCYIIAAACYIIACACQIIAGCiT0/DdpXMU6qZhu2+H8ltLVwmBbtMjk52LseeyzHkcjy4TPeWgpt831dxBgQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMBE6k1E7ONchmm6DiMNal+u/blwGVqZnu7/9ziXAaYuXIdpBrXm48ePD2Q/+/btC2Q/rlyOIVdBHXtfBWdAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATDCMFM5cBl0GNeTSdbhjMg8jdVnvzs5O3zWuBgwYEEiNyzqcP3/ed43k9nUKqiYVcAYEADBBAAEATPgOoJ07d2rOnDkqLCxUWlqaNm/eHHe/53lauXKlCgoKNHDgQJWVlenIkSOJ6hcAkCJ8B1BLS4tKSkq0du3aq96/Zs0avfLKK3r99de1e/duDR48WOXl5Wptbb3hZgEAqcP3RQgVFRWqqKi46n2e5+nll1/W888/r7lz50qS3njjDeXn52vz5s166KGHbqxbAEDKSOhrQPX19WpsbFRZWVnstnA4rNLSUtXW1l61pq2tTdFoNG4DAKS+hAZQY2OjJCk/Pz/u9vz8/Nh9X1ZVVaVwOBzbioqKEtkSACBJmV8Ft2LFCjU3N8e2EydOWLcEAAhAQgMoEolIkpqamuJub2pqit33ZaFQSNnZ2XEbACD1JTSAiouLFYlEtH379tht0WhUu3fv1rRp0xK5KwBAL+f7Krjz58/r6NGjsY/r6+t14MAB5eTkaPjw4Vq2bJl+/vOf6/bbb1dxcbFeeOEFFRYWat68eYnsGwDQy/kOoD179ui+++6Lfbx8+XJJ0sKFC7V+/Xo9++yzamlp0RNPPKGzZ8/q7rvv1rZt25xmPgEAUlea5zLZrwdFo1GFw2GNGzfOaTCkH65P3WXAY1CDJC9duuS7pqOjw3eNJLW3twdS4/Kckn0YaVCDXF3WznVfEyZM8F1z5513+q5xeVP7jh07fNdIUmZmpu+afv38z3h2qXH9+RjEsNTOzk4dPnxYzc3N13xd3/wqOABA30QAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMOF/BCucph+7TBcOiutUcJfn5LKvIKb33kidyzq41AQ1HV2SCgoKfNeMHTvWd43L91JDQ4PvGtfjIRUl01pwBgQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEw0hTjMuwT9dhpKkoqCGhLjUdHR2+a4YNG+a7RpKmTJniVOfX1q1bfde4rF1mZqbvmiD11e9BzoAAACYIIACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYIIAAgCY6NPDSNPS0qxbuKagBosm+yDEUCjkuyY3N9dpXy0tLb5rLl686Lvm0qVLvmtuueUW3zUTJkzwXePq/PnzvmtcBou6CPIYT/bvp2TCGRAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATfXoYaZBScbDowIEDfdd861vf8l3jMoz0pptu8l0jJfcwUpf1zsjI8F0jSenp/n83/eSTT3zXuKxDZmam7xr8l8vPiJ4a3MwZEADABAEEADDhO4B27typOXPmqLCwUGlpadq8eXPc/YsWLVJaWlrcNnv27ET1CwBIEb4DqKWlRSUlJVq7dm23j5k9e7YaGhpi28aNG2+oSQBA6vF9EUJFRYUqKiqu+ZhQKKRIJOLcFAAg9fXIa0DV1dXKy8vTmDFjtGTJEp05c6bbx7a1tSkajcZtAIDUl/AAmj17tt544w1t375dv/zlL1VTU6OKiopu/+/3qqoqhcPh2FZUVJTolgAASSjh7wN66KGHYv++4447NGnSJI0aNUrV1dWaOXPmFY9fsWKFli9fHvs4Go0SQgDQB/T4ZdgjR45Ubm6ujh49etX7Q6GQsrOz4zYAQOrr8QA6efKkzpw5o4KCgp7eFQCgF/H9J7jz58/Hnc3U19frwIEDysnJUU5OjlavXq0FCxYoEono2LFjevbZZ3XbbbepvLw8oY0DAHo33wG0Z88e3XfffbGPv3j9ZuHChXrttdd08OBB/f73v9fZs2dVWFioWbNm6Wc/+5nTPC8AQOryHUAzZsy45jC7P/3pTzfUUG8Q5MDPILj+cjBjxgzfNYMGDfJd090VlNfS1dXlu0ZyG/jZr5//a3lchnC6cBkq6lrnMrCyp4Zcflmqfc9K7se464DansAsOACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYIIAAgCYIIAAACYIIACAiYT/l9y9SZATcl325TLtduLEib5rRo4c6btGcpsc3d7e7rvGZTKzywRtKbmnJre0tPiucV2HnJwc3zXf+c53fNf87W9/813z2Wef+a4Jaup20Pvq7TgDAgCYIIAAACYIIACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYKJPDyN15TIk1KWmuLjYd83s2bN91/znP//xXSNJra2tvmtcBosGNchVkvr18/8t4bKvffv2+a45cuSI7xpXw4YN810zfvx43zWTJ0/2XZOdne275vDhw75rpOAGi/bVAaacAQEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADDBMNIkVl5e7rvm0qVLvms6Ozt917ju67PPPvNdE4lEfNe4DiM9c+aM75qamhrfNS0tLb5rXIayujp16pTvGpehtoMHD/ZdU1pa6rtm9OjRvmskqb6+3ndNUINFXQb7Jpve/wwAAL0SAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwwjDcg999zju+b222/3XeMy7NN1GGlHR4fvmiFDhviuaWho8F3T2trqu0aS/vrXv/quOX/+vO8a1zX3y3VgpctATZcBsG1tbb5r9u7d67tmwIABvmskKSMjw3eNy5oHNcA02XAGBAAwQQABAEz4CqCqqipNmTJFWVlZysvL07x581RXVxf3mNbWVlVWVmro0KG66aabtGDBAjU1NSW0aQBA7+crgGpqalRZWaldu3bpgw8+UEdHh2bNmhX3n2s9/fTTev/99/Xuu++qpqZGp06d0gMPPJDwxgEAvZuvixC2bdsW9/H69euVl5envXv3avr06WpubtZvf/tbbdiwQffff78kad26dRo3bpx27dqlb37zm4nrHADQq93Qa0DNzc2SpJycHEmXr07p6OhQWVlZ7DFjx47V8OHDVVtbe9XP0dbWpmg0GrcBAFKfcwB1dXVp2bJluuuuuzRx4kRJUmNjo/r373/Fpbb5+flqbGy86uepqqpSOByObUVFRa4tAQB6EecAqqys1KFDh/TWW2/dUAMrVqxQc3NzbDtx4sQNfT4AQO/g9EbUpUuXauvWrdq5c6eGDRsWuz0Siai9vV1nz56NOwtqampSJBK56ucKhUIKhUIubQAAejFfZ0Ce52np0qXatGmTduzYoeLi4rj7J0+erMzMTG3fvj12W11dnY4fP65p06YlpmMAQErwdQZUWVmpDRs2aMuWLcrKyoq9rhMOhzVw4ECFw2E9/vjjWr58uXJycpSdna2nnnpK06ZN4wo4AEAcXwH02muvSZJmzJgRd/u6deu0aNEiSdKvf/1rpaena8GCBWpra1N5ebl+85vfJKRZAEDqSPM8z7Nu4n9Fo1GFw2GNGzfOaRCgH65P3WWQ5MaNG33XXLx40XfN/v37fdecOnXKd42kbq9svJaPP/7Yd83Jkyd917gO+7x06VJg+/LLZchlZmam07769+8fSI1Lfy41/fq5zV12WXOXn1suw0hdfz4GMfi0s7NThw8fVnNzs7Kzs7t9HLPgAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYIIAAgCYIIAAACYIIACACQIIAGCCAAIAmCCAAAAm3EbE9nEuE3L/7//+z3fNypUrfdcUFBT4rjl48KDvGkl6//33fdd0dHT4rglieu+N7Cuo6ccuE52DnIYd1GRrl+8/lxopuK+ta3+9Xd981gAAcwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEz06WGkQQ653Lt3r++aOXPm+K7p7Oz0XXPp0iXfNZL7oMsguA53dFk/Fy5DLl1qXL9GQQ0+dXlOLl9bl/1IwQ0WddlPkD+/egpnQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEz06WGkQXIdjulXkEMNXeqCGqjZ1dXlu0aSPM9zqvMrqCGXrkM4g9pXUMeQ6zHOYNGexRkQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAwQQABAEwQQAAAEwwjdRDU4ECXoYudnZ2+a/r1czsMXAY1ugwJdRkQ6jpU1HWIqV+pOOTS5TkFNaTXdT99dUhoUDgDAgCYIIAAACZ8BVBVVZWmTJmirKws5eXlad68eaqrq4t7zIwZM5SWlha3PfnkkwltGgDQ+/kKoJqaGlVWVmrXrl364IMP1NHRoVmzZqmlpSXucYsXL1ZDQ0NsW7NmTUKbBgD0fr5efd62bVvcx+vXr1deXp727t2r6dOnx24fNGiQIpFIYjoEAKSkG3oNqLm5WZKUk5MTd/ubb76p3NxcTZw4UStWrNCFCxe6/RxtbW2KRqNxGwAg9Tlfht3V1aVly5bprrvu0sSJE2O3P/LIIxoxYoQKCwt18OBBPffcc6qrq9N777131c9TVVWl1atXu7YBAOil0jzHN0wsWbJEf/zjH/Xxxx9r2LBh3T5ux44dmjlzpo4ePapRo0ZdcX9bW5va2tpiH0ejURUVFWncuHFO74NJVq7vS/HL5X1ArlzeM8P7gC7jfUDuNS6S/X1AqfZ+o87OTh0+fFjNzc3Kzs7u9nFOZ0BLly7V1q1btXPnzmuGjySVlpZKUrcBFAqFFAqFXNoAAPRivgLI8zw99dRT2rRpk6qrq1VcXHzdmgMHDkiSCgoKnBoEAKQmXwFUWVmpDRs2aMuWLcrKylJjY6MkKRwOa+DAgTp27Jg2bNigb3/72xo6dKgOHjyop59+WtOnT9ekSZN65AkAAHonX68Bdfd3ynXr1mnRokU6ceKEvv/97+vQoUNqaWlRUVGR5s+fr+eff/6afwf8X9FoVOFwmNeAHPEakHuNxGtAN4LXgNzxGtBXcL1v6qKiItXU1Pj5lACAPopp2AFx+Q3H5bf4oH6jdOXSX5BnQMksyN+Sg9pXUMer6/NJtTOTZJPcP60AACmLAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYIIAAgCYIIAAACYIIACACYaRJrFkH7/PoMbLknnwabJ/jZK9P/QszoAAACYIIACACQIIAGCCAAIAmCCAAAAmCCAAgAkCCABgggACAJgggAAAJgggAIAJAggAYCLpZsF9MVers7PTuBNcTzLPQAtSMq9Dss9aS/b+4OaLn9/X+95IugA6d+6cJOnTTz817gQAcCPOnTuncDjc7f1pXpL9+tbV1aVTp04pKyvrit+OotGoioqKdOLECWVnZxt1aI91uIx1uIx1uIx1uCwZ1sHzPJ07d06FhYVKT+/+lZ6kOwNKT0/XsGHDrvmY7OzsPn2AfYF1uIx1uIx1uIx1uMx6Ha515vMFLkIAAJgggAAAJnpVAIVCIa1atUqhUMi6FVOsw2Wsw2Wsw2Wsw2W9aR2S7iIEAEDf0KvOgAAAqYMAAgCYIIAAACYIIACAiV4TQGvXrtWtt96qAQMGqLS0VH/5y1+sWwrciy++qLS0tLht7Nix1m31uJ07d2rOnDkqLCxUWlqaNm/eHHe/53lauXKlCgoKNHDgQJWVlenIkSM2zfag663DokWLrjg+Zs+ebdNsD6mqqtKUKVOUlZWlvLw8zZs3T3V1dXGPaW1tVWVlpYYOHaqbbrpJCxYsUFNTk1HHPeOrrMOMGTOuOB6efPJJo46vrlcE0Ntvv63ly5dr1apV2rdvn0pKSlReXq7Tp09btxa4CRMmqKGhIbZ9/PHH1i31uJaWFpWUlGjt2rVXvX/NmjV65ZVX9Prrr2v37t0aPHiwysvL1draGnCnPet66yBJs2fPjjs+Nm7cGGCHPa+mpkaVlZXatWuXPvjgA3V0dGjWrFlqaWmJPebpp5/W+++/r3fffVc1NTU6deqUHnjgAcOuE++rrIMkLV68OO54WLNmjVHH3fB6galTp3qVlZWxjzs7O73CwkKvqqrKsKvgrVq1yispKbFuw5Qkb9OmTbGPu7q6vEgk4r300kux286ePeuFQiFv48aNBh0G48vr4Hmet3DhQm/u3Lkm/Vg5ffq0J8mrqanxPO/y1z4zM9N79913Y485fPiwJ8mrra21arPHfXkdPM/z7r33Xu+HP/yhXVNfQdKfAbW3t2vv3r0qKyuL3Zaenq6ysjLV1tYadmbjyJEjKiws1MiRI/Xoo4/q+PHj1i2Zqq+vV2NjY9zxEQ6HVVpa2iePj+rqauXl5WnMmDFasmSJzpw5Y91Sj2pubpYk5eTkSJL27t2rjo6OuONh7NixGj58eEofD19ehy+8+eabys3N1cSJE7VixQpduHDBor1uJd0w0i/7/PPP1dnZqfz8/Ljb8/Pz9Y9//MOoKxulpaVav369xowZo4aGBq1evVr33HOPDh06pKysLOv2TDQ2NkrSVY+PL+7rK2bPnq0HHnhAxcXFOnbsmH7yk5+ooqJCtbW1ysjIsG4v4bq6urRs2TLdddddmjhxoqTLx0P//v01ZMiQuMem8vFwtXWQpEceeUQjRoxQYWGhDh48qOeee051dXV67733DLuNl/QBhP+qqKiI/XvSpEkqLS3ViBEj9M477+jxxx837AzJ4KGHHor9+4477tCkSZM0atQoVVdXa+bMmYad9YzKykodOnSoT7wOei3drcMTTzwR+/cdd9yhgoICzZw5U8eOHdOoUaOCbvOqkv5PcLm5ucrIyLjiKpampiZFIhGjrpLDkCFDNHr0aB09etS6FTNfHAMcH1caOXKkcnNzU/L4WLp0qbZu3aqPPvoo7r9viUQiam9v19mzZ+Men6rHQ3frcDWlpaWSlFTHQ9IHUP/+/TV58mRt3749dltXV5e2b9+uadOmGXZm7/z58zp27JgKCgqsWzFTXFysSCQSd3xEo1Ht3r27zx8fJ0+e1JkzZ1Lq+PA8T0uXLtWmTZu0Y8cOFRcXx90/efJkZWZmxh0PdXV1On78eEodD9dbh6s5cOCAJCXX8WB9FcRX8dZbb3mhUMhbv3699/e//9174oknvCFDhniNjY3WrQXqRz/6kVddXe3V19d7n3zyiVdWVubl5uZ6p0+ftm6tR507d87bv3+/t3//fk+S96tf/crbv3+/969//cvzPM/7xS9+4Q0ZMsTbsmWLd/DgQW/u3LlecXGxd/HiRePOE+ta63Du3DnvmWee8Wpra736+nrvww8/9L7+9a97t99+u9fa2mrdesIsWbLEC4fDXnV1tdfQ0BDbLly4EHvMk08+6Q0fPtzbsWOHt2fPHm/atGnetGnTDLtOvOutw9GjR72f/vSn3p49e7z6+npvy5Yt3siRI73p06cbdx6vVwSQ53neq6++6g0fPtzr37+/N3XqVG/Xrl3WLQXuwQcf9AoKCrz+/ft7t9xyi/fggw96R48etW6rx3300UeepCu2hQsXep53+VLsF154wcvPz/dCoZA3c+ZMr66uzrbpHnCtdbhw4YI3a9Ys7+abb/YyMzO9ESNGeIsXL065X9Ku9vwleevWrYs95uLFi94PfvAD72tf+5o3aNAgb/78+V5DQ4Nd0z3geutw/Phxb/r06V5OTo4XCoW82267zfvxj3/sNTc32zb+Jfx3DAAAE0n/GhAAIDURQAAAEwQQAMAEAQQAMEEAAQBMEEAAABMEEADABAEEADBBAAEATBBAAAATBBAAwAQBBAAw8f/6BMT+bo/1rgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "random_nr = np.random.randint(0, 56000) #Selects random image from traning set. There are 56,000 images in traning set.\n",
    "\n",
    "#Printing out the 2D array of the image\n",
    "plt.imshow(X_train[random_nr].reshape(28,28), cmap=\"gray\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b29df949",
   "metadata": {},
   "source": [
    "Now that we are certain the dataset is valid and ready to be used, we can now begin image augmentation. Image augmentation artificially increases the size of a dataset by modifying images by rotating and zooming them. This is done to prevent overfitting, and to make the model more robust. The images are rotated by 10 degrees, and zoomed by 10%. The images are then reshaped back into a 28 by 28 matrix."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ff672ecb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(56000, 28, 28, 1) (56000,)\n",
      "(64, 28, 28, 1) (64,)\n",
      "(9984, 28, 28, 1)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Attempting to use Keras's image augmentation\n",
    "import keras.preprocessing.image\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "print(X_train.shape, y_train.shape)\n",
    "datagen = ImageDataGenerator(\n",
    "        rotation_range=25,  \n",
    "        zoom_range = 0.20,  \n",
    "        )\n",
    "datagen.fit(X_train)\n",
    "\n",
    "gen_augmented = datagen.flow(X_train, y_train, batch_size=64)\n",
    "augmented_images = []\n",
    "augmented_labels = []\n",
    "for i in range(10000 // 64):\n",
    "    batch_augmented_images, batch_augmented_labels = next(gen_augmented)\n",
    "    augmented_images.append(batch_augmented_images)\n",
    "    augmented_labels.append(batch_augmented_labels)\n",
    "\n",
    "augmented_images = np.concatenate(augmented_images, axis=0)\n",
    "augmented_labels = np.concatenate(augmented_labels, axis=0)\n",
    "\n",
    "\n",
    "print(batch_augmented_images.shape, batch_augmented_labels.shape)\n",
    "print(augmented_images.shape)\n",
    "fig, axs = plt.subplots(3, 3, figsize=(10, 10))\n",
    "for i, ax in enumerate(axs.flatten()):\n",
    "    ax.imshow(augmented_images[i].reshape(28, 28), cmap=\"gray\")\n",
    "    ax.set_title(f\"Label: {augmented_labels[i]}\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec352ea9",
   "metadata": {},
   "source": [
    "We can now apply CNN model to train and evaluate our dataset. We use multiple layers shown below, and use accuracy and MAE to evaluate the model. Overall, CNN performs well."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f560a990",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1031/1031 [==============================] - 46s 43ms/step - loss: 0.1643 - accuracy: 0.9487 - mae: 0.0160 - val_loss: 0.0630 - val_accuracy: 0.9805 - val_mae: 0.0066\n",
      "Epoch 2/5\n",
      "1031/1031 [==============================] - 37s 36ms/step - loss: 0.0523 - accuracy: 0.9831 - mae: 0.0052 - val_loss: 0.0505 - val_accuracy: 0.9849 - val_mae: 0.0042\n",
      "Epoch 3/5\n",
      "1031/1031 [==============================] - 39s 38ms/step - loss: 0.0371 - accuracy: 0.9880 - mae: 0.0036 - val_loss: 0.0355 - val_accuracy: 0.9892 - val_mae: 0.0030\n",
      "Epoch 4/5\n",
      "1031/1031 [==============================] - 42s 41ms/step - loss: 0.0280 - accuracy: 0.9913 - mae: 0.0028 - val_loss: 0.0416 - val_accuracy: 0.9876 - val_mae: 0.0032\n",
      "Epoch 5/5\n",
      "1031/1031 [==============================] - 42s 41ms/step - loss: 0.0229 - accuracy: 0.9926 - mae: 0.0023 - val_loss: 0.0348 - val_accuracy: 0.9899 - val_mae: 0.0027\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import to_categorical\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense\n",
    "# One-hot encode the labels\n",
    "y_train_onehot = to_categorical(y_train)\n",
    "y_test_onehot = to_categorical(y_test)\n",
    "\n",
    "# Define the model architecture\n",
    "model = Sequential()\n",
    "model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))\n",
    "model.add(MaxPooling2D((2, 2)))\n",
    "model.add(Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(MaxPooling2D((2, 2)))\n",
    "model.add(Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "# Compile the model\n",
    "model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=[\"accuracy\",\"mae\", ])\n",
    "\n",
    "# Train the model\n",
    "history = model.fit(X_train, y_train_onehot, epochs=5, batch_size=64, validation_data=(X_test, y_test_onehot))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "2012a312",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the training and validation loss curves\n",
    "plt.plot(history.history['loss'], label='training loss')\n",
    "plt.plot(history.history['val_loss'], label='validation loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "# Plot the training and validation accuracy curves\n",
    "plt.plot(history.history['accuracy'], label='training accuracy')\n",
    "plt.plot(history.history['val_accuracy'], label='validation accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.ylabel('accuracy')\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "ac787bb9",
   "metadata": {},
   "source": [
    "The code below tests out different numbers of augmented images.  four different models are trained, using 1000, 5000, 10000, and 20000 extra images from image augmentation.  The x_train dataset size was increased by 2%, 9%, 18%, and 36% respectively.  The accuracy of the models were all above 99%, but surprisingly the image augmentation had a slight negative effect on the accuracy and MAE of the model.  The MAE was always below .01, and was slowly raised as the number of augmented images increased.  The cause of this is most likely that the model was overfitting with the augmented images, and it negatively affected the accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "2806ac73",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(56960, 28, 28, 1) (56960,)\n",
      "(960, 28, 28, 1)\n",
      "Epoch 1/5\n",
      "890/890 [==============================] - 35s 37ms/step - loss: 0.1787 - accuracy: 0.9461 - mae: 0.0172 - val_loss: 0.0566 - val_accuracy: 0.9830 - val_mae: 0.0058\n",
      "Epoch 2/5\n",
      "890/890 [==============================] - 34s 38ms/step - loss: 0.0593 - accuracy: 0.9814 - mae: 0.0057 - val_loss: 0.0451 - val_accuracy: 0.9862 - val_mae: 0.0038\n",
      "Epoch 3/5\n",
      "890/890 [==============================] - 38s 42ms/step - loss: 0.0411 - accuracy: 0.9871 - mae: 0.0040 - val_loss: 0.0355 - val_accuracy: 0.9882 - val_mae: 0.0033\n",
      "Epoch 4/5\n",
      "890/890 [==============================] - 32s 36ms/step - loss: 0.0331 - accuracy: 0.9894 - mae: 0.0032 - val_loss: 0.0351 - val_accuracy: 0.9890 - val_mae: 0.0031\n",
      "Epoch 5/5\n",
      "890/890 [==============================] - 32s 35ms/step - loss: 0.0265 - accuracy: 0.9911 - mae: 0.0026 - val_loss: 0.0439 - val_accuracy: 0.9884 - val_mae: 0.0030\n",
      "(60992, 28, 28, 1) (60992,)\n",
      "(4992, 28, 28, 1)\n",
      "Epoch 1/5\n",
      "953/953 [==============================] - 36s 36ms/step - loss: 0.2302 - accuracy: 0.9302 - mae: 0.0215 - val_loss: 0.0557 - val_accuracy: 0.9831 - val_mae: 0.0053\n",
      "Epoch 2/5\n",
      "953/953 [==============================] - 34s 35ms/step - loss: 0.0792 - accuracy: 0.9747 - mae: 0.0077 - val_loss: 0.0538 - val_accuracy: 0.9843 - val_mae: 0.0049\n",
      "Epoch 3/5\n",
      "953/953 [==============================] - 35s 36ms/step - loss: 0.0522 - accuracy: 0.9837 - mae: 0.0051 - val_loss: 0.0470 - val_accuracy: 0.9876 - val_mae: 0.0035\n",
      "Epoch 4/5\n",
      "953/953 [==============================] - 34s 35ms/step - loss: 0.0400 - accuracy: 0.9875 - mae: 0.0040 - val_loss: 0.0448 - val_accuracy: 0.9901 - val_mae: 0.0029\n",
      "Epoch 5/5\n",
      "953/953 [==============================] - 34s 36ms/step - loss: 0.0309 - accuracy: 0.9901 - mae: 0.0031 - val_loss: 0.0485 - val_accuracy: 0.9899 - val_mae: 0.0027\n",
      "(65984, 28, 28, 1) (65984,)\n",
      "(9984, 28, 28, 1)\n",
      "Epoch 1/5\n",
      "1031/1031 [==============================] - 39s 36ms/step - loss: 0.2676 - accuracy: 0.9188 - mae: 0.0251 - val_loss: 0.0547 - val_accuracy: 0.9836 - val_mae: 0.0052\n",
      "Epoch 2/5\n",
      "1031/1031 [==============================] - 38s 37ms/step - loss: 0.0920 - accuracy: 0.9720 - mae: 0.0089 - val_loss: 0.0513 - val_accuracy: 0.9845 - val_mae: 0.0046\n",
      "Epoch 3/5\n",
      "1031/1031 [==============================] - 37s 36ms/step - loss: 0.0625 - accuracy: 0.9805 - mae: 0.0061 - val_loss: 0.0497 - val_accuracy: 0.9846 - val_mae: 0.0040\n",
      "Epoch 4/5\n",
      "1031/1031 [==============================] - 42s 41ms/step - loss: 0.0463 - accuracy: 0.9854 - mae: 0.0046 - val_loss: 0.0459 - val_accuracy: 0.9876 - val_mae: 0.0031\n",
      "Epoch 5/5\n",
      "1031/1031 [==============================] - 67s 65ms/step - loss: 0.0382 - accuracy: 0.9880 - mae: 0.0037 - val_loss: 0.0409 - val_accuracy: 0.9906 - val_mae: 0.0025\n",
      "(75968, 28, 28, 1) (75968,)\n",
      "(19968, 28, 28, 1)\n",
      "Epoch 1/5\n",
      "1187/1187 [==============================] - 45s 36ms/step - loss: 0.2836 - accuracy: 0.9107 - mae: 0.0270 - val_loss: 0.0679 - val_accuracy: 0.9797 - val_mae: 0.0058\n",
      "Epoch 2/5\n",
      "1187/1187 [==============================] - 42s 36ms/step - loss: 0.0946 - accuracy: 0.9701 - mae: 0.0093 - val_loss: 0.0439 - val_accuracy: 0.9874 - val_mae: 0.0035\n",
      "Epoch 3/5\n",
      "1187/1187 [==============================] - 43s 36ms/step - loss: 0.0614 - accuracy: 0.9805 - mae: 0.0061 - val_loss: 0.0376 - val_accuracy: 0.9892 - val_mae: 0.0031\n",
      "Epoch 4/5\n",
      "1187/1187 [==============================] - 43s 36ms/step - loss: 0.0468 - accuracy: 0.9850 - mae: 0.0047 - val_loss: 0.0483 - val_accuracy: 0.9864 - val_mae: 0.0036\n",
      "Epoch 5/5\n",
      "1187/1187 [==============================] - 43s 36ms/step - loss: 0.0356 - accuracy: 0.9888 - mae: 0.0036 - val_loss: 0.0452 - val_accuracy: 0.9899 - val_mae: 0.0025\n"
     ]
    }
   ],
   "source": [
    "additionalImages=[1000, 5000, 10000, 20000]\n",
    "\n",
    "for iteration in additionalImages:\n",
    "    datagen = ImageDataGenerator(\n",
    "        rotation_range=20,  \n",
    "        zoom_range = 0.15,\n",
    "        width_shift_range=0.2,\n",
    "        brightness_range=[0.5,1.0],\n",
    "          \n",
    "        )\n",
    "    datagen.fit(X_train)\n",
    "\n",
    "    gen_augmented = datagen.flow(X_train, y_train, batch_size=64)\n",
    "    augmented_images = []\n",
    "    augmented_labels = []\n",
    "    for i in range(iteration // 64):\n",
    "        batch_augmented_images, batch_augmented_labels = next(gen_augmented)\n",
    "        augmented_images.append(batch_augmented_images)\n",
    "        augmented_labels.append(batch_augmented_labels)\n",
    "    augmented_images = np.concatenate(augmented_images, axis=0)\n",
    "    augmented_labels = np.concatenate(augmented_labels, axis=0)\n",
    "    X_train_augmented = np.concatenate([X_train, augmented_images], axis=0)\n",
    "    y_train_augmented = np.concatenate([y_train, augmented_labels], axis=0)\n",
    "\n",
    "    print(X_train_augmented.shape, y_train_augmented.shape)\n",
    "    print(augmented_images.shape)\n",
    "    y_train_onehot = to_categorical(y_train_augmented)\n",
    "    y_test_onehot = to_categorical(y_test)\n",
    "\n",
    "    # Define the model architecture\n",
    "    model = Sequential()\n",
    "    model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1)))\n",
    "    model.add(MaxPooling2D((2, 2)))\n",
    "    model.add(Conv2D(64, (3, 3), activation='relu'))\n",
    "    model.add(MaxPooling2D((2, 2)))\n",
    "    model.add(Conv2D(64, (3, 3), activation='relu'))\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(64, activation='relu'))\n",
    "    model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "    # Compile the model\n",
    "    model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=[\"accuracy\",\"mae\", ])\n",
    "\n",
    "    # Train the model\n",
    "    history = model.fit(X_train_augmented, y_train_onehot, epochs=5, batch_size=64, validation_data=(X_test, y_test_onehot))\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fb2ba944",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}