doc/example_usage.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# HOAG usage example\n",
    "\n",
    "We will load an example dataset (20news from scikit-learn) and learn with HOAG the optimal L2 reagularization parameter."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    }
   ],
   "source": [
    "%pylab inline\n",
    "\n",
    "from hoag import LogisticRegressionCV\n",
    "\n",
    "# load some data\n",
    "from sklearn import datasets, linear_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get a training set and test set\n",
    "data_train = datasets.fetch_20newsgroups_vectorized(subset='train')\n",
    "data_test = datasets.fetch_20newsgroups_vectorized(subset='test')\n",
    "\n",
    "X_train = data_train.data\n",
    "X_test = data_test.data\n",
    "y_train = data_train.target\n",
    "y_test = data_test.target\n",
    "\n",
    "# binarize labels\n",
    "y_train[data_train.target < 10] = -1\n",
    "y_train[data_train.target >= 10] = 1\n",
    "y_test[data_test.target < 10] = -1\n",
    "y_test[data_test.target >= 10] = 1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegressionCV()"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# optimize model parameters and hyperparameters jointly\n",
    "# using HOAG\n",
    "clf = LogisticRegressionCV()\n",
    "clf.fit(X_train, y_train, X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Regularization chosen by HOAG: alpha=-4.5625978805564005\n"
     ]
    }
   ],
   "source": [
    "print('Regularization chosen by HOAG: alpha=%s' % (clf.alpha_[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So above is the the optimal regularization parameter chose by HOAG.\n",
    "\n",
    "Lets now plot the cost function to see how well HOAG did.  Note that HOAG minimizes a parametrization in which ```clf.alpha_``` is inside an exponential hence we need to take this into account when comparing with the scikit-learn estimator. \n",
    "\n",
    "WARNING: this might take a long time to compute (~10 min)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n",
      "/home/zaccharie/workspace/hoag/venv/lib/python3.6/site-packages/sklearn/linear_model/_logistic.py:765: ConvergenceWarning: lbfgs failed to converge (status=1):\n",
      "STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.\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",
      "Please also refer to the documentation for alternative solver options:\n",
      "    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression\n",
      "  extra_warning_msg=_LOGISTIC_SOLVER_CONVERGENCE_MSG)\n"
     ]
    }
   ],
   "source": [
    "# range for regularization parameters\n",
    "alphas = np.linspace(-25, 10, 40)\n",
    "\n",
    "def cost_func(a):\n",
    "    clf = linear_model.LogisticRegression(\n",
    "        solver='lbfgs',\n",
    "        C=np.exp(-a), fit_intercept=False, \n",
    "        tol=1e-22, max_iter=500)\n",
    "\n",
    "    clf.fit(X_train, y_train)\n",
    "    cost = linear_model._logistic._logistic_loss(clf.coef_.ravel(), X_test, y_test, 0.)\n",
    "    return cost\n",
    "\n",
    "scores = [cost_func(a) for a in alphas]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# make the plot bigger than default\n",
    "plt.rcParams['figure.figsize'] = (8.0, 6.0)\n",
    "plt.rcParams['font.size'] = 20\n",
    "\n",
    "# plot the scores\n",
    "plt.plot(alphas, scores, lw=3, label='cross-validation error')\n",
    "plt.xlabel(r'$\\alpha$', fontsize=40)\n",
    "\n",
    "# plot HOAG value of alpha\n",
    "plt.plot((clf.alpha_[0], clf.alpha_[0]), (1000, 7000), c='k', linestyle='--', \n",
    "         label=r'value of regularization ($\\alpha$) found by HOAG')\n",
    "plt.legend(fontsize=20)\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "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": 1
}