23-logistic-regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Logistic regression with SGD\n",
    "\n",
    "This notebook contains a simple linear classifier trained with SGD."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/thead/anaconda/envs/toolbox-talk/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import SGDClassifier\n",
    "\n",
    "import json\n",
    "import numpy as np\n",
    "\n",
    "import utils"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data loading and augmentation\n",
    "\n",
    "We **do not** augment the dataset by adding rotations and flips of the original images to the training data.\n",
    "\n",
    "For our SGD estimator we need to linearize the data from a 3D tensor to a vector."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "X, y = utils.load_data(one_hot=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def linearize(X):\n",
    "    return X.reshape(X.shape[0], 80 * 80 * 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=1)\n",
    "\n",
    "def augment(X, y):\n",
    "    X_prime = np.concatenate((\n",
    "        X,\n",
    "        np.rot90(X, 1, (1,2)),\n",
    "        np.rot90(X, 2, (1,2)),\n",
    "        np.rot90(X, 3, (1,2)),\n",
    "        np.flip(X, 1),\n",
    "        np.flip(X, 1)\n",
    "        ))\n",
    "    y_prime = np.concatenate((y, y, y, y, y, y))\n",
    "    return X_prime, y_prime\n",
    "\n",
    "#X_train, y_train = augment(X_train, y_train)\n",
    "\n",
    "X_train = linearize(X_train)\n",
    "X_test = linearize(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((2100, 19200), (2100,))"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape, y_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup, train and score classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "tags": [
     "parameters"
    ]
   },
   "outputs": [],
   "source": [
    "alpha = 0.0001\n",
    "penalty = 'l2'\n",
    "loss = 'hinge'\n",
    "max_iter = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "clf = SGDClassifier(max_iter=max_iter, loss=loss, penalty=penalty, alpha=alpha)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(alpha=0.0001, average=False, class_weight=None, epsilon=0.1,\n",
       "       eta0=0.0, fit_intercept=True, l1_ratio=0.15,\n",
       "       learning_rate='optimal', loss='hinge', max_iter=50, n_iter=None,\n",
       "       n_jobs=1, penalty='l2', power_t=0.5, random_state=None,\n",
       "       shuffle=True, tol=None, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy = clf.score(X_test, y_test)\n",
    "train_accuracy = clf.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test accuracy: 0.9171428571428571\n"
     ]
    }
   ],
   "source": [
    "print(\"Test accuracy:\", accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import papermill as pm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/papermill.record+json": {
       "alpha": 0.0001
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/papermill.record+json": {
       "loss": "hinge"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/papermill.record+json": {
       "penalty": "l2"
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/papermill.record+json": {
       "max_iter": 50
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/papermill.record+json": {
       "test_accuracy": 0.9171428571428571
      }
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/papermill.record+json": {
       "train_accuracy": 0.9685714285714285
      }
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pm.record('alpha', alpha)\n",
    "pm.record('loss', loss)\n",
    "pm.record('penalty', penalty)\n",
    "pm.record('max_iter', max_iter)\n",
    "pm.record('test_accuracy', accuracy)\n",
    "pm.record('train_accuracy', train_accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "celltoolbar": "Tags",
  "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}