Support Vector Machine-6-Stochastic Gradient Descent-Nonlinear Data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Support Vector Machine: SGD on Nonlinear Data\n",
    "\n",
    "In this notebook, we use the Stochastic Gradient Descent (SGD) algorithm for the SVM model to solve a binary classification problem on a **linearly non-separable** dataset.\n",
    "\n",
    "Unlike the kernel SVM (implemented by Scikit-Learn's SVC class), the SGD algorithm for SVM (implemented by Scikit-Learn's SGDClassifier) is unable to separate nonlinear data. First we need to augment the features by adding polynomial terms. \n",
    "\n",
    "We experiment with two cases:\n",
    "\n",
    "- Case 1: No Feature Augmentation\n",
    "- Case 2: Feature Augmentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.datasets import make_circles\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler, PolynomialFeatures\n",
    "from sklearn.linear_model import SGDClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Function to Draw Decision Boundary "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def decision_boundary_class_colored(clf, X, plotDistanceFromHyperplane=False, colorBar=False):\n",
    "    \n",
    "    # Get the min and max value of feature x1\n",
    "    x1min, x1max = X[:,0].min() - 1, X[:, 0].max() + 1\n",
    "    \n",
    "    # Get the min and max value of feature x2\n",
    "    x2min, x2max = X[:,1].min() - 1, X[:, 1].max() + 1\n",
    "    \n",
    "    # Create the mesh grid\n",
    "    x1s = np.linspace(x1min, x1max, 100)\n",
    "    x2s = np.linspace(x2min, x2max, 100)\n",
    "    x1, x2 = np.meshgrid(x1s, x2s)\n",
    "    \n",
    "    \n",
    "    # Create pairs of new points from the grid\n",
    "    X_new = np.c_[x1.ravel(), x2.ravel()]\n",
    "    \n",
    "    \n",
    "    # Compute the class predictions for all new points\n",
    "    y_pred = clf.predict(X_new).reshape(x1.shape)\n",
    "    \n",
    "    \n",
    "    # Generate the contourf plot for the predictions\n",
    "    #plt.contourf(x1, x2, y_pred, cmap=plt.cm.bwr, alpha=0.2)\n",
    "    plt.contourf(x1, x2, y_pred, cmap=plt.cm.RdGy, alpha=0.99)\n",
    "    \n",
    "    \n",
    "    if(plotDistanceFromHyperplane == True):\n",
    "    \n",
    "        # Compute the signed distance of a sample to the hyperplane for all new points\n",
    "        y_decision = clf.decision_function(X_new).reshape(x1.shape)\n",
    "\n",
    "        # Generate the contour plot for the distance of all points from the hyperplane and the two margins\n",
    "        plt.contour(x1, x2, y_decision, levels=[-1, 0, 1], alpha=0.5, linestyles=['--', '-', '--'], colors='black')\n",
    "        \n",
    "        \n",
    "        #plt.pcolormesh(x1, x2, -y_decision, cmap=plt.cm.RdBu)\n",
    "        \n",
    "        # Generate the contourf plot for the distance of all points from the hyperplane\n",
    "        plt.contourf(x1, x2, y_decision, cmap=plt.cm.bwr, alpha=0.2)\n",
    "    \n",
    "    \n",
    "    if(colorBar==True):\n",
    "        plt.colorbar()   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Linearly Non-Separable Dataset\n",
    "\n",
    "We use Scikit-Learn's \"datasets.make_circles\" function to generate a synthetic nonlinear dataset that is suitable for investigating classification (as well as clustering) models. It creates a large circle containing a smaller circle in 2d.\n",
    "\n",
    "The following two parameters are used by the make_circles function:\n",
    "\n",
    "factor : Scale factor between inner and outer circle. Range 0 ~ 1.\n",
    "\n",
    "noise : Standard deviation of Gaussian noise added to the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "X, y = make_circles(300, factor=0.4, noise=0.15)\n",
    "label = y\n",
    "\n",
    "plt.figure(figsize=(12, 6))\n",
    "\n",
    "plt.plot(X[y.ravel()==0, 0], X[y.ravel()==0, 1], \"wo\", markeredgecolor=\"k\")\n",
    "plt.plot(X[y.ravel()==1, 0], X[y.ravel()==1, 1], \"yo\")\n",
    "\n",
    "plt.title(\"Linearly Non-separable Dataset\", fontsize=16)\n",
    "plt.xlabel(\"$x_1$\", fontsize=14)\n",
    "plt.ylabel(\"$x_2$\", fontsize=14)\n",
    "plt.xticks(())\n",
    "plt.yticks(())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create Train & Test Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SVM using SGD: No Feature Augmentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 432 candidates, totalling 1296 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  34 tasks      | elapsed:    1.3s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Score (F1 micro): 0.735584\n",
      "Optimal Hyperparameter Values:  {'svm_sgd__alpha': 0.01, 'svm_sgd__eta0': 0.1, 'svm_sgd__learning_rate': 'constant', 'svm_sgd__max_iter': 500, 'svm_sgd__penalty': 'l2', 'svm_sgd__tol': 0.001}\n",
      "\n",
      "\n",
      "CPU times: user 1.59 s, sys: 115 ms, total: 1.71 s\n",
      "Wall time: 2.98 s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done 1296 out of 1296 | elapsed:    2.9s finished\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "svm_sgd_pipeline = Pipeline([\n",
    "        (\"scaler\", StandardScaler()),\n",
    "        (\"svm_sgd\", SGDClassifier(loss='hinge')),\n",
    "    ])\n",
    "\n",
    "\n",
    "param_grid = {'svm_sgd__alpha': [0.01, 0.001],\n",
    "              'svm_sgd__penalty' : [\"l2\", \"l1\"],\n",
    "              'svm_sgd__learning_rate': [\"constant\", \"optimal\", \"invscaling\", \"adaptive\"], \n",
    "              'svm_sgd__max_iter':[500, 1000, 3000],\n",
    "              'svm_sgd__eta0': [0.1, 0.01, 0.001],\n",
    "              'svm_sgd__tol': [1e-3, 1e-5, 1e-8]}\n",
    "\n",
    "\n",
    "svm_clf = GridSearchCV(svm_sgd_pipeline, param_grid, scoring='f1', cv=3, verbose=1, n_jobs=-1)\n",
    "svm_clf.fit(X_train, y_train)\n",
    "\n",
    "params_optimal = svm_clf.best_params_\n",
    "\n",
    "print(\"Best Score (F1 micro): %f\" % svm_clf.best_score_)\n",
    "print(\"Optimal Hyperparameter Values: \", params_optimal)\n",
    "print(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## No Feature Augmentation: Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Train Accuracy:  0.4875\n",
      "Test Accuracy:  0.5333333333333333\n"
     ]
    }
   ],
   "source": [
    "y_train_predicted_svm = svm_clf.predict(X_train)\n",
    "\n",
    "accuracy_score_train_svm = np.mean(y_train_predicted_svm == y_train)\n",
    "print(\"\\nTrain Accuracy: \", accuracy_score_train_svm)\n",
    "\n",
    "y_test_predicted_svm = svm_clf.predict(X_test)\n",
    "\n",
    "accuracy_score_test_svm = np.mean(y_test_predicted_svm == y_test)\n",
    "print(\"Test Accuracy: \", accuracy_score_test_svm)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Polynomial SVM using SGD: Feature Augmentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 1296 candidates, totalling 3888 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done 268 tasks      | elapsed:    0.6s\n",
      "[Parallel(n_jobs=-1)]: Done 2368 tasks      | elapsed:    3.7s\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Score (F1 micro): 0.979721\n",
      "Optimal Hyperparameter Values:  {'poly__degree': 2, 'svm_sgd__alpha': 0.01, 'svm_sgd__eta0': 0.1, 'svm_sgd__learning_rate': 'constant', 'svm_sgd__max_iter': 1000, 'svm_sgd__penalty': 'l1', 'svm_sgd__tol': 0.001}\n",
      "\n",
      "\n",
      "CPU times: user 5.87 s, sys: 89.6 ms, total: 5.96 s\n",
      "Wall time: 6.52 s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Done 3888 out of 3888 | elapsed:    6.5s finished\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "# Create a Pipeline object\n",
    "svm_sgd_polynomial_pipeline = Pipeline([\n",
    "        # Bias should be excluded because by default SGDRegressor adds bias via the \"fit_intercept\" parameter\n",
    "        ('poly', PolynomialFeatures(include_bias=False)), \n",
    "        ('scaler', StandardScaler()),\n",
    "        (\"svm_sgd\", SGDClassifier(loss='hinge')),\n",
    "    ])\n",
    "\n",
    "param_grid = {'poly__degree': [1, 2, 3],\n",
    "              'svm_sgd__alpha': [0.01, 0.001],\n",
    "              'svm_sgd__penalty' : [\"l2\", \"l1\"],\n",
    "              'svm_sgd__learning_rate': [\"constant\", \"optimal\", \"invscaling\", \"adaptive\"], \n",
    "              'svm_sgd__max_iter':[500, 1000, 3000],\n",
    "              'svm_sgd__eta0': [0.1, 0.01, 0.001],\n",
    "              'svm_sgd__tol': [1e-3, 1e-5, 1e-8]}\n",
    "\n",
    "\n",
    "svm_clf_polynomial = GridSearchCV(svm_sgd_polynomial_pipeline, param_grid, scoring='f1', cv=3, verbose=1, n_jobs=-1)\n",
    "svm_clf_polynomial.fit(X_train, y_train)\n",
    "\n",
    "params_optimal = svm_clf_polynomial.best_params_\n",
    "\n",
    "print(\"Best Score (F1 micro): %f\" % svm_clf_polynomial.best_score_)\n",
    "print(\"Optimal Hyperparameter Values: \", params_optimal)\n",
    "print(\"\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature Augmentation: Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Train Accuracy:  0.9791666666666666\n",
      "Test Accuracy:  0.9666666666666667\n"
     ]
    }
   ],
   "source": [
    "y_train_predicted_svm_polynomial = svm_clf_polynomial.predict(X_train)\n",
    "\n",
    "accuracy_score_train_svm_polynomial = np.mean(y_train_predicted_svm_polynomial == y_train)\n",
    "print(\"\\nTrain Accuracy: \", accuracy_score_train_svm_polynomial)\n",
    "\n",
    "y_test_predicted_svm_polynomial = svm_clf_polynomial.predict(X_test)\n",
    "\n",
    "accuracy_score_test_svm_polynomial = np.mean(y_test_predicted_svm_polynomial == y_test)\n",
    "print(\"Test Accuracy: \", accuracy_score_test_svm_polynomial)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compare Decision Boundaries: Linear SVM vs Polynomial SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(18, 6))\n",
    "plt.subplot(121)\n",
    "\n",
    "plt.plot(X[y.ravel()==0, 0], X[y.ravel()==0, 1], \"wo\", markeredgecolor=\"k\", label=\"Class 0\")\n",
    "plt.plot(X[y.ravel()==1, 0], X[y.ravel()==1, 1], \"yo\", markeredgecolor=\"k\", label=\"Class 1\")\n",
    "\n",
    "decision_boundary_class_colored(svm_clf, X, plotDistanceFromHyperplane=False, colorBar=False)\n",
    "\n",
    "plt.xlabel(\"x1\", fontsize=14)\n",
    "plt.ylabel(\"x2\", fontsize=14)\n",
    "plt.legend(loc=\"best\", fontsize=14)\n",
    "plt.title(\"Linear SVM: Test Accuracy %.2f\" % accuracy_score_test_svm)\n",
    "plt.xticks(())\n",
    "plt.yticks(())\n",
    "\n",
    "\n",
    "plt.subplot(122)\n",
    "plt.plot(X[y.ravel()==0, 0], X[y.ravel()==0, 1], \"wo\", markeredgecolor=\"k\")\n",
    "plt.plot(X[y.ravel()==1, 0], X[y.ravel()==1, 1], \"yo\", markeredgecolor=\"k\")\n",
    "\n",
    "decision_boundary_class_colored(svm_clf_polynomial, X, plotDistanceFromHyperplane=False, colorBar=False)\n",
    "\n",
    "plt.xlabel(\"x1\", fontsize=14)\n",
    "plt.ylabel(\"x2\", fontsize=14)\n",
    "plt.title(\"Polynomial SVM: Test Accuracy %.2f\" % accuracy_score_test_svm_polynomial)\n",
    "plt.xticks(())\n",
    "plt.yticks(())\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}