simple_simulated_demo.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# L2 Regularized Logistic Regression Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This demo file provides an example of using my L2 Regularized Logistic Regression model built with numpy to classify some simple, simulated data.\n",
    "\n",
    "- **Author:** Joel Stremmel (jstremme@uw.edu)\n",
    "- **Credits:** University of Washington DATA 558 with Zaid Harchaoui and Corinne Jones"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Standard Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.metrics import precision_recall_fscore_support as score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import Methods for Modeling Training and Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from l2_regularized_logistic_regression import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Generate Some Simulated Data and Targets with Signal and Noise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x1 mean: 5.011148067568902, x1 sd: 0.9969531724556675\n",
      "x2 mean: 2.0104214753322958, x2 sd: 0.9731896374278008\n"
     ]
    }
   ],
   "source": [
    "x1 = 5 * np.ones(shape=(50, 80)) + np.random.normal(size=(50, 80))\n",
    "x2 = 2 * np.ones(shape=(50, 80)) + np.random.normal(size=(50, 80))\n",
    "\n",
    "mu1, sd1 = np.mean(x1), np.std(x1)\n",
    "mu2, sd2 = np.mean(x2), np.std(x2)\n",
    "\n",
    "print(\"x1 mean: {}, x1 sd: {}\".format(mu1, sd1))\n",
    "print(\"x2 mean: {}, x2 sd: {}\".format(mu2, sd2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = np.concatenate((x1, x2), axis=0)\n",
    "y = np.concatenate((np.ones(50), np.ones(50) * -1), axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train Test Split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=0, shuffle=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Scale Input Data and Transform Binary Targets to [-1, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_scaler = StandardScaler()\n",
    "X_train = X_scaler.fit_transform(X_train)\n",
    "X_test = X_scaler.transform(X_test)\n",
    "y_train = transform_target(y_train)\n",
    "y_test = transform_target(y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train L2 Regularized Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "beta_vals = l2_log_reg(X_train, y_train, lambda_penalty=1.0, eps=0.001, v=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize Value of Object from Each Iteration of Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_objective_values(X_train, y_train, beta_vals, lambda_penalty=1.0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict with Each Coefficient Vector on Train and Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_pred_vals = predict_all(beta_vals, X_train)\n",
    "test_pred_vals = predict_all(beta_vals, X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize Misclassification Error at Each Step of Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_errors(train_pred_vals, test_pred_vals, y_train, y_test, labels=['Train Set', 'Test Set'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use K Fold Cross Validation to Find the Optimal Lambda Value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Lambda=0.001 mean misclassification error on test sets: 0.0\n",
      "Lambda=0.01 mean misclassification error on test sets: 0.0\n",
      "Lambda=0.1 mean misclassification error on test sets: 0.0\n",
      "Lambda=1 mean misclassification error on test sets: 0.0\n",
      "Lambda=10 mean misclassification error on test sets: 0.0\n",
      "Lambda=100 mean misclassification error on test sets: 0.0\n",
      "Lambda=1000 mean misclassification error on test sets: 0.0\n"
     ]
    }
   ],
   "source": [
    "lambdas = [0.001, 0.01, 0.1, 1, 10, 100, 1000]\n",
    "mean_test_errors = k_fold_crossval(X_train, y_train, lambdas, k=3, eps=0.001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best lambda: 0.001\n"
     ]
    }
   ],
   "source": [
    "best_lambda = lambdas[np.argmin(mean_test_errors)]\n",
    "print('Best lambda: {}'.format(best_lambda))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Retrain L2 Regularized Logistic Regression with Optimal Lambda"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "beta_vals = l2_log_reg(X_train, y_train, lambda_penalty=best_lambda, eps=0.001, v=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict with Each Coefficient Vector on Train and Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_pred_vals = predict_all(beta_vals, X_train)\n",
    "test_pred_vals = predict_all(beta_vals, X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualize Misclassification Error at Each Step of Gradient Descent"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_errors(train_pred_vals, test_pred_vals, y_train, y_test, labels=['Train Set', 'Test Set'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculate Precision and Recall with the Final Beta from Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Class</th>\n",
       "      <th>TrainPrecision</th>\n",
       "      <th>TrainRecall</th>\n",
       "      <th>ValidationPrecision</th>\n",
       "      <th>ValidationRecall</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Class  TrainPrecision  TrainRecall  ValidationPrecision  ValidationRecall\n",
       "0      0             1.0          1.0                  1.0               1.0\n",
       "1      1             1.0          1.0                  1.0               1.0"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_precision, train_recall, train_fscore, train_support = score(y_train, train_pred_vals[-1])\n",
    "val_precision, val_recall, val_fscore, val_support = score(y_test, test_pred_vals[-1])\n",
    "\n",
    "pr_df = pd.DataFrame({'Class': list(range(0, 2)),\n",
    "                      'TrainPrecision': train_precision,\n",
    "                      'TrainRecall': train_recall,\n",
    "                      'ValidationPrecision': val_precision,\n",
    "                      'ValidationRecall': val_recall})\n",
    "pr_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}