第8章 逻辑回归算法.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:25.151775Z",
     "start_time": "2018-11-06T15:09:21.494932Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.datasets import load_iris\n",
    "from sklearn.model_selection import cross_val_score, train_test_split\n",
    "from sklearn.feature_extraction.text import CountVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:25.182856Z",
     "start_time": "2018-11-06T15:09:25.155200Z"
    }
   },
   "outputs": [],
   "source": [
    "iris = load_iris()\n",
    "X = iris.data[:,:2]\n",
    "y = iris.target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:25.503028Z",
     "start_time": "2018-11-06T15:09:25.187295Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lr = LogisticRegression(C=1e5, multi_class='ovr', solver='lbfgs')\n",
    "lr.fit(X,y)\n",
    "x1_min, x1_max = X[:,0].min(), X[:,0].max()\n",
    "x2_min, x2_max = X[:,1].min(), X[:,1].max()\n",
    "xx1, xx2 = np.meshgrid(np.linspace(x1_min,x1_max), np.linspace(x2_min,x2_max))\n",
    "Z = lr.predict(np.c_[xx1.ravel(), xx2.ravel()])\n",
    "Z = Z.reshape(xx1.shape)\n",
    "plt.figure(1, figsize=(8,6))\n",
    "plt.pcolormesh(xx1, xx2, Z, cmap=plt.cm.Paired)\n",
    "plt.scatter(X[:,0], X[:,1], c=y, edgecolors='k', cmap=plt.cm.Paired)\n",
    "plt.xlabel('Sepal length')\n",
    "plt.ylabel('Sepal width')\n",
    "plt.xlim(xx1.min(), xx1.max())\n",
    "plt.ylim(xx2.min(), xx2.max())\n",
    "plt.xticks(())\n",
    "plt.yticks(())\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:25.518528Z",
     "start_time": "2018-11-06T15:09:25.510661Z"
    }
   },
   "outputs": [],
   "source": [
    "import glob\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:25.544193Z",
     "start_time": "2018-11-06T15:09:25.524146Z"
    }
   },
   "outputs": [],
   "source": [
    "def features_and_label(path, train=True, attack=False):\n",
    "    if train:\n",
    "        fileList = glob.glob(path + '*.txt')\n",
    "        featuresList_train = []\n",
    "        for file in fileList:\n",
    "            featuresList_train.append(' '.join(np.loadtxt(file).astype('int').astype('str')))\n",
    "        label_train = [0] * len(fileList)\n",
    "        return featuresList_train, label_train\n",
    "    if attack:\n",
    "        fileList = glob.glob(path + '*.txt')\n",
    "        featuresList_attack = []\n",
    "        for file in fileList:\n",
    "            featuresList_attack.append(' '.join(np.loadtxt(file).astype('int').astype('str'))) \n",
    "        label_attack = [1] * len(fileList)\n",
    "        return featuresList_attack, label_attack"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:30.424964Z",
     "start_time": "2018-11-06T15:09:25.549591Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9625"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "normal, normal_label = features_and_label('./数据集/ADFA-LD/Training_Data_Master/')\n",
    "attack, attack_label = features_and_label('./数据集/ADFA-LD/Attack_Data_Master/Java_Meterpreter_*/', train=False, attack=True)\n",
    "\n",
    "countvec = CountVectorizer()\n",
    "X = countvec.fit_transform(normal + attack)\n",
    "y = normal_label + attack_label\n",
    "\n",
    "X_train, X_val, y_train, y_val = train_test_split(X, y)\n",
    "\n",
    "lr = LogisticRegression(solver='sag', C=1e5, max_iter=10000)\n",
    "lr.fit(X_train, y_train)\n",
    "lr.score(X_val, y_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:43.419849Z",
     "start_time": "2018-11-06T15:09:30.429381Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9602720880810274"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cross_val_score(lr, X, y, cv=10).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:09:47.738466Z",
     "start_time": "2018-11-06T15:09:43.423769Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-11-06T15:10:03.524616Z",
     "start_time": "2018-11-06T15:10:03.023949Z"
    }
   },
   "outputs": [],
   "source": [
    "mnist_train, mnist_test = mnist.load_data()\n",
    "X_train = mnist_train[0].reshape(60000, 28 * 28)\n",
    "X_test = mnist_test[0].reshape(10000, 28 * 28)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2018-11-06T15:10:11.511Z"
    }
   },
   "outputs": [],
   "source": [
    "lr = LogisticRegression(solver='saga', C=1e5, max_iter=10000, multi_class='ovr')\n",
    "lr.fit(X_train, mnist_train[1])\n",
    "lr.score(X_test, mnist_test[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2018-11-06T15:10:13.118Z"
    }
   },
   "outputs": [],
   "source": [
    "cross_val_score(lr, X_train, mnist_train[1], cv=10, scoring='accuracy').mean()"
   ]
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}