kdd99-svm.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#读取训练数据\n",
    "train_data = pd.read_csv(\"kddcup_train_data\", header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#读取测试数据\n",
    "test_data = pd.read_csv(\"kddcup_test_data\", header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#定义数据每一列的列名\n",
    "col_names = [\"duration\",\"protocol_type\",\"service\",\"flag\",\"src_bytes\",\"dst_bytes\",\"land\",\"wrong_fragment\",\n",
    "             \"urgent\",\"hot\",\"num_failed_logins\",\"logged_in\",\"num_compromised\",\n",
    "             \"root_shell\",\"su_attempted\",\"num_root\",\"num_file_creations\",\n",
    "             \"num_shells\",\"num_access_files\",\"num_outbound_cmds\",\n",
    "             \"is_hot_login\",\"is_guest_login\",\"count\",\"srv_count\",\"serror_rate\",\n",
    "             \"srv_serror_rate\",\"rerror_rate\",\"srv_rerror_rate\",\"same_srv_rate\",\n",
    "             \"diff_srv_rate\",\"srv_diff_host_rate\",\"dst_host_count\",\"dst_host_srv_count\",\n",
    "             \"dst_host_same_srv_rate\",\"dst_host_diff_srv_rate\",\"dst_host_same_src_port_rate\",\n",
    "             \"dst_host_srv_diff_host_rate\",\"dst_host_serror_rate\",\"dst_host_srv_serror_rate\",\n",
    "             \"dst_host_rerror_rate\",\"dst_host_srv_rerror_rate\",\"target\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#添加列名\n",
    "train_data.columns = col_names\n",
    "test_data.columns = col_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#把训练数据和测试数据拆分为: 训练数据特征（x_train）训练数据标签（y_train）测试数据特征（x_test）测试数据标签（y_test）\n",
    "x_train = train_data.iloc[:,:train_data.shape[1]-1]\n",
    "y_train = train_data.iloc[:,train_data.shape[1]-1:]\n",
    "\n",
    "x_test = test_data.iloc[:,:test_data.shape[1]-1]\n",
    "y_test = test_data.iloc[:,test_data.shape[1]-1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#将三个特征为单词的列做 \"one hot\" 转换\n",
    "x_train_dummy = pd.get_dummies(x_train, columns=[\"protocol_type\",\"service\",\"flag\"])\n",
    "x_test_dummy = pd.get_dummies(x_test, columns=[\"protocol_type\",\"service\",\"flag\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "service_red_i\n",
      "service_urh_i\n"
     ]
    },
    {
     "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>duration</th>\n",
       "      <th>src_bytes</th>\n",
       "      <th>dst_bytes</th>\n",
       "      <th>land</th>\n",
       "      <th>wrong_fragment</th>\n",
       "      <th>urgent</th>\n",
       "      <th>hot</th>\n",
       "      <th>num_failed_logins</th>\n",
       "      <th>logged_in</th>\n",
       "      <th>num_compromised</th>\n",
       "      <th>...</th>\n",
       "      <th>flag_REJ</th>\n",
       "      <th>flag_RSTO</th>\n",
       "      <th>flag_RSTOS0</th>\n",
       "      <th>flag_RSTR</th>\n",
       "      <th>flag_S0</th>\n",
       "      <th>flag_S1</th>\n",
       "      <th>flag_S2</th>\n",
       "      <th>flag_S3</th>\n",
       "      <th>flag_SF</th>\n",
       "      <th>flag_SH</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>105</td>\n",
       "      <td>146</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1 rows × 118 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   duration  src_bytes  dst_bytes  land  wrong_fragment  urgent  hot  \\\n",
       "0         0        105        146     0               0       0    0   \n",
       "\n",
       "   num_failed_logins  logged_in  num_compromised  ...  flag_REJ  flag_RSTO  \\\n",
       "0                  0          0                0  ...         0          0   \n",
       "\n",
       "   flag_RSTOS0  flag_RSTR  flag_S0  flag_S1  flag_S2  flag_S3  flag_SF  \\\n",
       "0            0          0        0        0        0        0        1   \n",
       "\n",
       "   flag_SH  \n",
       "0        0  \n",
       "\n",
       "[1 rows x 118 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 检查训练和测试数据集的列数是否一致\n",
    "missing_cols = set(x_train_dummy.columns) - set(x_test_dummy.columns)\n",
    "for c in missing_cols:\n",
    "    print(c)\n",
    "    x_test_dummy[c] = 0\n",
    "    \n",
    "# 确保列的顺序是一致的\n",
    "x_test_dummy = x_test_dummy[x_train_dummy.columns]\n",
    "\n",
    "from IPython.display import display\n",
    "display(x_test_dummy.head(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/ec2-user/anaconda3/envs/pytorch_p27/lib/python2.7/site-packages/sklearn/preprocessing/data.py:334: DataConversionWarning: Data with input dtype uint8, int64, float64 were all converted to float64 by MinMaxScaler.\n",
      "  return self.partial_fit(X, y)\n"
     ]
    }
   ],
   "source": [
    "# 为了SVM可以收敛，将特征值修改到-1到1之间\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "x_train_scale = x_train_dummy\n",
    "scaling = MinMaxScaler(feature_range=(-1,1)).fit(x_train_scale)\n",
    "x_train_scale = scaling.transform(x_train_scale)\n",
    "                                             \n",
    "x_test_scale = x_test_dummy\n",
    "x_test_scale = scaling.transform(x_test_scale)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LibSVM]model completed, using time 276.65 seconds\n"
     ]
    }
   ],
   "source": [
    "#训练并生成模型\n",
    "\n",
    "import time\n",
    "from sklearn import svm\n",
    "rseed = 93\n",
    "time_start = time.time()\n",
    "model_svm = svm.SVC(kernel='linear', C=1, verbose=True, random_state=rseed, decision_function_shape=\"ovo\").fit(x_train_scale, y_train.values.ravel())\n",
    "print(\"training completed, using time %5.2f seconds\" % (time.time()-time_start))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['model_svm.sav']"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#保存模型到本地\n",
    "\n",
    "from sklearn.externals import joblib\n",
    "model_name = \"model_svm.sav\"\n",
    "joblib.dump(model_svm,model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy based on training: 0.9987\n",
      "accuracy based on testing: 0.9240\n"
     ]
    }
   ],
   "source": [
    "# 计算预测的准确率，分别用训练数据集和测试数据集\n",
    "print(\"accuracy based on training: %5.4f\" % model_svm.score(x_train_scale, y_train))  \n",
    "\n",
    "print(\"accuracy based on testing: %5.4f\" % model_svm.score(x_test_scale, y_test)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 计算 Confusion Matrix 并生成图形\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from matplotlib import pyplot as plt\n",
    "import itertools\n",
    "\n",
    "cnf_model_svm = confusion_matrix(model_svm.predict(x_test_scale),y_test)\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    np.set_printoptions(precision=2)\n",
    "\n",
    "\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        #print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print(\"\")\n",
    "        #print('Confusion matrix, without normalization')\n",
    "\n",
    "    #print(cm)\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True Value')\n",
    "    plt.xlabel('Predicted Value')\n",
    "    \n",
    "\n",
    "import numpy as np\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "width = 6\n",
    "height = 6\n",
    "plt.figure(figsize=(width, height))\n",
    "plot_confusion_matrix(cnf_model_svm, classes=model_svm.classes_, title=\"Confusion Matrix without Normalization\")\n",
    "\n",
    "plt.figure(figsize=(width, height))\n",
    "plot_confusion_matrix(cnf_model_svm, classes=model_svm.classes_, normalize=True, title=\"Confusion Matrix with Normalization\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "conda_pytorch_p27",
   "language": "python",
   "name": "conda_pytorch_p27"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}