09_01_linear_svm.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:01.663565Z",
     "start_time": "2017-06-22T08:35:01.245471Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:01.955691Z",
     "start_time": "2017-06-22T08:35:01.665001Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_breast_cancer\n",
    "data = load_breast_cancer()\n",
    "X = data.data\n",
    "y = data.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:01.986452Z",
     "start_time": "2017-06-22T08:35:01.956745Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import ShuffleSplit\n",
    "ss = ShuffleSplit(n_splits=1, \n",
    "                  train_size=0.8, \n",
    "                  test_size=0.2, \n",
    "                  random_state=0)\n",
    "\n",
    "train_index, test_index = next(ss.split(X, y))\n",
    "\n",
    "X_train, X_test = X[train_index], X[test_index]\n",
    "y_train, y_test = y[train_index], y[test_index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:01.992369Z",
     "start_time": "2017-06-22T08:35:01.987936Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "scaler = MinMaxScaler()\n",
    "scaler.fit(X_train)\n",
    "X_train = scaler.transform(X_train)\n",
    "X_test  = scaler.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:01.996123Z",
     "start_time": "2017-06-22T08:35:01.994127Z"
    }
   },
   "outputs": [],
   "source": [
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:02.039730Z",
     "start_time": "2017-06-22T08:35:01.997479Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "\n",
    "clf = SVC(kernel='linear')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:02.050684Z",
     "start_time": "2017-06-22T08:35:02.040918Z"
    }
   },
   "outputs": [],
   "source": [
    "clf.C = 1\n",
    "\n",
    "st = time.time()\n",
    "clf.fit(X_train, y_train)\n",
    "print(1000 * (time.time() - st), \"[ms]\")\n",
    "\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:02.245921Z",
     "start_time": "2017-06-22T08:35:02.051591Z"
    }
   },
   "outputs": [],
   "source": [
    "clf.C = 1e20\n",
    "\n",
    "st = time.time()\n",
    "clf.fit(X_train, y_train)\n",
    "print(1000 * (time.time() - st), \"[ms]\")\n",
    "\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:03.877603Z",
     "start_time": "2017-06-22T08:35:02.247067Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "param = {'C': 10**np.arange(-15.0,21.0)}\n",
    "\n",
    "gs = GridSearchCV(clf, param, verbose=1, return_train_score=True)\n",
    "gs.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:04.389471Z",
     "start_time": "2017-06-22T08:35:03.878671Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.plot(gs.cv_results_['param_C'].data, \n",
    "         gs.cv_results_['mean_fit_time'], \n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs.cv_results_['param_C'].data,\n",
    "         gs.cv_results_['mean_score_time'], \n",
    "         label=\"test(val)\")\n",
    "\n",
    "plt.ylim(0,)\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.ylabel(\"computation time [sec]\")\n",
    "plt.legend(loc=\"best\");\n",
    "plt.title(\"SVM with linear kernel\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:04.393238Z",
     "start_time": "2017-06-22T08:35:04.390693Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.svm import LinearSVC\n",
    "\n",
    "clf = LinearSVC()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:04.403106Z",
     "start_time": "2017-06-22T08:35:04.394849Z"
    }
   },
   "outputs": [],
   "source": [
    "clf.C = 1\n",
    "\n",
    "st = time.time()\n",
    "clf.fit(X_train, y_train)\n",
    "print(1000 * (time.time() - st), \"[ms]\")\n",
    "\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 警告を非表示にする（追加コード）\n",
    "import warnings\n",
    "from sklearn.exceptions import ConvergenceWarning\n",
    "warnings.simplefilter('ignore', ConvergenceWarning)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:04.417084Z",
     "start_time": "2017-06-22T08:35:04.404465Z"
    }
   },
   "outputs": [],
   "source": [
    "clf.C = 1e20\n",
    "\n",
    "st = time.time()\n",
    "clf.fit(X_train, y_train)\n",
    "print(1000 * (time.time() - st), \"[ms]\")\n",
    "\n",
    "clf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:04.781187Z",
     "start_time": "2017-06-22T08:35:04.418397Z"
    },
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "param = {'C': 10**np.arange(-15.0,21.0)}\n",
    "\n",
    "gs2 = GridSearchCV(clf, param, verbose=1, return_train_score=True)\n",
    "gs2.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:05.136684Z",
     "start_time": "2017-06-22T08:35:04.782313Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.plot(gs2.cv_results_['param_C'].data, \n",
    "         gs2.cv_results_['mean_fit_time'], \n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs2.cv_results_['param_C'].data,\n",
    "         gs2.cv_results_['mean_score_time'], \n",
    "         label=\"test(val)\")\n",
    "\n",
    "plt.ylim(0,)\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.ylabel(\"computation time [sec]\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.title(\"LinearSVM\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:05.429331Z",
     "start_time": "2017-06-22T08:35:05.137748Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.plot(gs.cv_results_['param_C'].data, \n",
    "         gs.cv_results_['mean_fit_time'], \n",
    "         label=\"training SVM (linear kernel)\")\n",
    "\n",
    "plt.plot(gs2.cv_results_['param_C'].data, \n",
    "         gs2.cv_results_['mean_fit_time'], \n",
    "         label=\"training LinearSVM\")\n",
    "\n",
    "plt.ylim(0,)\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.ylabel(\"computation time [sec]\")\n",
    "plt.legend(loc=\"best\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:05.436120Z",
     "start_time": "2017-06-22T08:35:05.430374Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "\n",
    "estimators = [('scaler', MinMaxScaler()), \n",
    "              ('clf', LinearSVC())]\n",
    "\n",
    "pipe = Pipeline(estimators)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:05.833356Z",
     "start_time": "2017-06-22T08:35:05.437206Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "param = {'clf__C': 10**np.arange(-15.0,21.0)}\n",
    "\n",
    "gs = GridSearchCV(pipe, param, verbose=1, return_train_score=True)\n",
    "gs.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:06.248121Z",
     "start_time": "2017-06-22T08:35:05.834579Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.plot(gs.cv_results_['param_clf__C'].data, \n",
    "         gs.cv_results_['mean_fit_time'], \n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data,\n",
    "         gs.cv_results_['mean_score_time'], \n",
    "         label=\"test(val)\")\n",
    "plt.ylabel(\"computation time [sec]\")\n",
    "plt.ylim(0,)\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.legend(loc=\"upper left\");\n",
    "\n",
    "plt.twinx()\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data, \n",
    "         gs.cv_results_['mean_train_score'], \n",
    "         linestyle=\"--\",\n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data,\n",
    "         gs.cv_results_['mean_test_score'], \n",
    "         linestyle=\"--\",\n",
    "         label=\"test(val)\")\n",
    "plt.ylabel(\"accuracy\")\n",
    "plt.legend(loc=\"lower right\");\n",
    "\n",
    "plt.title(\"LinearSVM\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:06.253549Z",
     "start_time": "2017-06-22T08:35:06.249500Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MaxAbsScaler\n",
    "\n",
    "estimators = [('scaler', MinMaxScaler()), \n",
    "              ('clf', SVC(kernel='linear'))]\n",
    "\n",
    "pipe = Pipeline(estimators)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:08.177143Z",
     "start_time": "2017-06-22T08:35:06.255150Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "param = {'clf__C': 10**np.arange(-15.0,21.0)}\n",
    "\n",
    "gs = GridSearchCV(pipe, param, verbose=1, return_train_score=True)\n",
    "gs.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:08.591170Z",
     "start_time": "2017-06-22T08:35:08.178354Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.plot(gs.cv_results_['param_clf__C'].data, \n",
    "         gs.cv_results_['mean_fit_time'], \n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data,\n",
    "         gs.cv_results_['mean_score_time'], \n",
    "         label=\"test(val)\")\n",
    "plt.ylabel(\"computation time [sec]\")\n",
    "plt.ylim(0,)\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.legend(loc=\"upper left\");\n",
    "\n",
    "plt.twinx()\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data, \n",
    "         gs.cv_results_['mean_train_score'], \n",
    "         linestyle=\"--\",\n",
    "         label=\"training\")\n",
    "\n",
    "plt.plot(gs.cv_results_['param_clf__C'].data,\n",
    "         gs.cv_results_['mean_test_score'], \n",
    "         linestyle=\"--\",\n",
    "         label=\"test(val)\")\n",
    "plt.ylabel(\"accuracy\")\n",
    "plt.legend(loc=\"lower right\");\n",
    "\n",
    "plt.title(\"SVM with linear kernel\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:08.953576Z",
     "start_time": "2017-06-22T08:35:08.592262Z"
    }
   },
   "outputs": [],
   "source": [
    "# from sklearn.datasets import fetch_mldata\n",
    "# mnist = fetch_mldata('MNIST original')\n",
    "from sklearn.datasets import fetch_openml\n",
    "mnist = fetch_openml('mnist_784', version=1)\n",
    "\n",
    "# MNISTの場合，60000が学習，10000がテスト，と決まっている\n",
    "# http://yann.lecun.com/exdb/mnist/\n",
    "\n",
    "X_train, X_test = mnist.data[:60000]/255., mnist.data[60000:]/255.\n",
    "y_train, y_test = mnist.target[:60000], mnist.target[60000:]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:08.958086Z",
     "start_time": "2017-06-22T08:35:08.954634Z"
    }
   },
   "outputs": [],
   "source": [
    "X_train.shape, X_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:08.961820Z",
     "start_time": "2017-06-22T08:35:08.959451Z"
    }
   },
   "outputs": [],
   "source": [
    "clf = SVC(kernel='linear')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:35:09.142663Z",
     "start_time": "2017-06-22T08:35:08.963253Z"
    }
   },
   "outputs": [],
   "source": [
    "st = time.time()\n",
    "clf.fit(X_train[::100], y_train[::100])\n",
    "print(1000 * (time.time() - st), \"[ms]\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:36:12.584523Z",
     "start_time": "2017-06-22T08:35:09.143738Z"
    }
   },
   "outputs": [],
   "source": [
    "for thin in [100, 50, 10, 5, 4]: # > 1 min...\n",
    "    st = time.time()\n",
    "    clf.fit(X_train[::thin], y_train[::thin])\n",
    "    et = time.time() - st\n",
    "    score = clf.score(X_test[::thin], y_test[::thin])\n",
    "    print(\"{0:.2f} sec, size {1}, accuracy {2}\".format(et, \n",
    "                                                       y_train[::thin].size,\n",
    "                                                       score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:36:12.592541Z",
     "start_time": "2017-06-22T08:36:12.586128Z"
    }
   },
   "outputs": [],
   "source": [
    "clf = LinearSVC() # dual"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:36:35.684962Z",
     "start_time": "2017-06-22T08:36:12.593985Z"
    }
   },
   "outputs": [],
   "source": [
    "for thin in [100, 50, 10, 5, 4]:\n",
    "    st = time.time()\n",
    "    clf.fit(X_train[::thin], y_train[::thin])\n",
    "    et = time.time() - st\n",
    "    score = clf.score(X_test[::thin], y_test[::thin])\n",
    "    print(\"{0:.2f} sec, size {1}, accuracy {2}\".format(et, \n",
    "                                                       y_train[::thin].size,\n",
    "                                                       score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:36:35.689233Z",
     "start_time": "2017-06-22T08:36:35.686506Z"
    }
   },
   "outputs": [],
   "source": [
    "clf = LinearSVC(dual=False) # primal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:36:45.802903Z",
     "start_time": "2017-06-22T08:36:35.691074Z"
    }
   },
   "outputs": [],
   "source": [
    "for thin in [100, 50, 10, 5, 4]:\n",
    "    st = time.time()\n",
    "    clf.fit(X_train[::thin], y_train[::thin])\n",
    "    et = time.time() - st\n",
    "    score = clf.score(X_test[::thin], y_test[::thin])\n",
    "    print(\"{0:.2f} sec, size {1}, accuracy {2}\".format(et, \n",
    "                                                       y_train[::thin].size,\n",
    "                                                       score))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T08:38:04.803113Z",
     "start_time": "2017-06-22T08:36:45.804653Z"
    }
   },
   "outputs": [],
   "source": [
    "for clf in [SVC(kernel='linear'), \n",
    "            LinearSVC(), \n",
    "            LinearSVC(dual=False)]:\n",
    "    times = []\n",
    "    sizes = []\n",
    "    for thin in [100, 50, 10, 5, 4]:\n",
    "        st = time.time()\n",
    "        clf.fit(X_train[::thin], y_train[::thin])\n",
    "        times.append(time.time() - st)\n",
    "        sizes.append(y_train[::thin].size)\n",
    "    plt.plot(sizes, times, label=clf.__class__.__name__)\n",
    "\n",
    "plt.legend(loc=\"best\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2017-06-22T09:00:37.978808Z",
     "start_time": "2017-06-22T08:38:04.804494Z"
    },
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "estimators = [('scaler', MinMaxScaler()), \n",
    "              ('clf', LinearSVC(dual=False))]\n",
    "\n",
    "pipe = Pipeline(estimators)\n",
    "\n",
    "param = {'clf__C': 10**np.arange(-5.0,10.0)}\n",
    "\n",
    "gs = GridSearchCV(pipe, param, n_jobs=-1, verbose=2, return_train_score=True)\n",
    "gs.fit(X_train, y_train) # about 30min..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2017-06-22T08:35:01.334Z"
    },
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "estimators = [('scaler', MinMaxScaler()), \n",
    "              ('clf', LinearSVC(dual=True))]\n",
    "\n",
    "pipe = Pipeline(estimators)\n",
    "\n",
    "param = {'clf__C': 10**np.arange(-5.0,10.0)}\n",
    "\n",
    "gs = GridSearchCV(pipe, param, n_jobs=-1, verbose=2, return_train_score=True)\n",
    "gs.fit(X_train, y_train) # 1 hour ?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2017-06-22T08:35:01.337Z"
    }
   },
   "outputs": [],
   "source": [
    "estimators = [('scaler', MinMaxScaler()), \n",
    "              ('clf', SVC(kernel='linear'))]\n",
    "\n",
    "pipe = Pipeline(estimators)\n",
    "\n",
    "param = {'clf__C': 10**np.arange(-5.0,10.0)}\n",
    "\n",
    "gs = GridSearchCV(pipe, param, n_jobs=-1, verbose=2, return_train_score=True)\n",
    "gs.fit(X_train, y_train) # 2 hours ?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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": "3.7.7"
  },
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "12px",
    "width": "252px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}