malware_detection.ipynb · 吴政琪/TLS-Malware-Detection-with-Machine-Learning

1月18日，北京，聊聊2025如何加入技术开发？

加入 Gitee

与超过 1200万开发者一起发现、参与优秀开源项目，私有仓库也完全免费：）

文件

该仓库未声明开源许可证文件（LICENSE），使用请关注具体项目描述及其代码上游依赖。

克隆/下载

malware_detection.ipynb 82.27 KB

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from math import sqrt;\n",
    "from sklearn import preprocessing\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.linear_model import LogisticRegression;\n",
    "from sklearn.metrics import accuracy_score, r2_score, confusion_matrix, mean_absolute_error, mean_squared_error, f1_score, log_loss\n",
    "from sklearn.model_selection import train_test_split\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns \n",
    "from sklearn.externals import joblib "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "False    62604\n",
      "dtype: int64\n",
      "False    16467\n",
      "dtype: int64\n",
      "Src_Port     0.0\n",
      "Dst_Port     0.0\n",
      "Bytes_in     0.0\n",
      "Bytes_out    0.0\n",
      "Pkts_in      0.0\n",
      "            ... \n",
      "sg_24        0.0\n",
      "sg_25        0.0\n",
      "ec_pts_0     0.0\n",
      "ec_pts_1     0.0\n",
      "isMalware    0.0\n",
      "Length: 439, dtype: float64\n",
      "0    55436\n",
      "1    14564\n",
      "Name: isMalware, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "#Importing Datasets \n",
    "malicious_dataset = pd.read_csv('insert malicious file name.csv')\n",
    "benign_dataset = pd.read_csv('insert benign file name.csv')\n",
    "\n",
    "# Removing duplicated rows from benign_dataset (5380 rows removed)\n",
    "benign_dataset = benign_dataset[benign_dataset.duplicated(keep=False) == False]\n",
    "\n",
    "print(benign_dataset.duplicated(keep=False).value_counts())\n",
    "print(malicious_dataset.duplicated(keep=False).value_counts())\n",
    "\n",
    "# Combining both datasets together\n",
    "all_flows = pd.concat([malicious_dataset, benign_dataset])\n",
    "\n",
    "#Inspecting datasets for columns and rows with missing values\n",
    "missing_values = all_flows.isnull().sum()\n",
    "overall_percentage = (missing_values/all_flows.isnull().count())\n",
    "print(overall_percentage)\n",
    "\n",
    "# Reducing the size of the dataset to reduce the amount of time taken in training models\n",
    "reduced_dataset = all_flows.sample(78000)\n",
    "\n",
    "validation_dataset = pd.read_csv('insert validation file name')\n",
    "validation_dataset = validation_dataset[validation_dataset.duplicated(keep=False) == False]\n",
    "\n",
    "# Examining the distribution of Malicious and Benign flows in the reduced dataset \n",
    "print(reduced_dataset['isMalware'].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Isolating independent and dependent variables for training dataset\n",
    "reduced_y = reduced_dataset['isMalware']\n",
    "reduced_x = reduced_dataset.drop(['isMalware'], axis=1);\n",
    "\n",
    "# Isolating independent and dependent variables for validation dataset\n",
    "validation_y = validation_dataset['isMalware']\n",
    "validation_x = validation_dataset.drop(['isMalware'], axis=1);\n",
    "\n",
    "# Splitting datasets into training and test data\n",
    "x_train, x_test, y_train, y_test = train_test_split(reduced_x, reduced_y, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[11065     0]\n",
      " [    4  2931]]\n",
      "Random Forest Classifier Accuracy score:  0.9997142857142857\n",
      "Random Forest Classifier with validation set Accuracy score:  0.9999551468939224\n"
     ]
    }
   ],
   "source": [
    "# Training random forest classifier\n",
    "rf_clf = RandomForestClassifier(max_depth=100)\n",
    "rf_clf.fit(x_train, y_train) \n",
    "rf_prediction = rf_clf.predict(x_test)\n",
    "conf_m = confusion_matrix(y_test, rf_prediction)  \n",
    "print(conf_m)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(y_test, rf_prediction))\n",
    "\n",
    "rf_validation_prediction = rf_clf.predict(validation_x)\n",
    "print('Random Forest Classifier with validation set Accuracy score: ', accuracy_score(rf_validation_prediction, validation_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean Absolute Error:  0.004857142857142857\n",
      "Root Mean Squared Error:  0.06969320524371696\n",
      "R Squared Score:  0.9706858419852952\n",
      "Accuracy score:  0.9951428571428571\n"
     ]
    }
   ],
   "source": [
    "# Training Logistics Regression classifier\n",
    "lm = LogisticRegression(max_iter=70000)\n",
    "lm.fit(x_train,y_train)\n",
    "predictions = lm.predict(x_test)\n",
    "rms = sqrt(mean_squared_error(y_test, predictions))\n",
    "print('Mean Absolute Error: ', mean_absolute_error(y_test, predictions))\n",
    "print('Root Mean Squared Error: ',rms)\n",
    "print('R Squared Score: ', r2_score(y_test, predictions))\n",
    "print('Accuracy score: ', accuracy_score(predictions, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean Absolute Error:  0.0033639829558196905\n",
      "Root Mean Squared Error:  0.05799985306722501\n",
      "R Squared Score:  0.0\n",
      "Accuracy score:  0.9966360170441803\n"
     ]
    }
   ],
   "source": [
    "#Testing the accuracy of the logistic regression model on the validation dataset\n",
    "validation_predictions = lm.predict(validation_x)\n",
    "rms = sqrt(mean_squared_error(validation_y, validation_predictions))\n",
    "print('Mean Absolute Error: ', mean_absolute_error(validation_y, validation_predictions))\n",
    "print('Root Mean Squared Error: ',rms)\n",
    "print('R Squared Score: ', r2_score(validation_y, validation_predictions))\n",
    "print('Accuracy score: ', accuracy_score(validation_predictions, validation_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[11039    26]\n",
      " [   42  2893]]\n"
     ]
    }
   ],
   "source": [
    "#Confusion matrix for test set accuracy\n",
    "conf_m_lr = confusion_matrix(y_test, predictions)  \n",
    "print(conf_m_lr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "False    16467\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Importing cipher stunted datasets \n",
    "cipher_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "\n",
    "# Removing duplicated rows from benign_dataset_test\n",
    "cipher_stunted_dataset = cipher_stunted_dataset[cipher_stunted_dataset.duplicated(keep=False) == False]\n",
    "\n",
    "print(cipher_stunted_dataset.duplicated(keep=False).value_counts())\n",
    "\n",
    "#Inspecting datasets for columns and rows with missing values\n",
    "test_missing_values = cipher_stunted_dataset.isnull().sum()\n",
    "test_overall_percentage = (test_missing_values/cipher_stunted_dataset.isnull().count())\n",
    "\n",
    "cs_test_x, cs_test_y = cipher_stunted_dataset.drop(['isMalware'], axis=1), cipher_stunted_dataset['isMalware']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "cs_test_predictions = lm.predict(cs_test_x)\n",
    "#print(test_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[    0     0]\n",
      " [16460     7]]\n"
     ]
    }
   ],
   "source": [
    "conf_m_lr_cs = confusion_matrix(cs_test_y, cs_test_predictions)  \n",
    "print(conf_m_lr_cs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistics Regression Accuracy score:  0.0004250926094613469\n"
     ]
    }
   ],
   "source": [
    "print('Logistics Regression Accuracy score: ', accuracy_score(cs_test_predictions, cs_test_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Classifier Accuracy score:  1.0\n"
     ]
    }
   ],
   "source": [
    "rf_validation_prediction_2 = rf_clf.predict(cs_test_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(rf_validation_prediction_2, cs_test_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[16467]]\n"
     ]
    }
   ],
   "source": [
    "conf_m_lr_cs = confusion_matrix(cs_test_y, rf_validation_prediction_2)  \n",
    "print(conf_m_lr_cs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Function to plot most important features of random forest model\n",
    "\n",
    "def plot_feature_importance(importance,names,model_type):\n",
    "\n",
    "    #Create arrays from feature importance and feature names\n",
    "    feature_importance = np.array(importance)\n",
    "    feature_names = np.array(names)\n",
    "\n",
    "    #Create a DataFrame using a Dictionary\n",
    "    data={'feature_names':feature_names,'feature_importance':feature_importance}\n",
    "    fi_df = pd.DataFrame(data)\n",
    "\n",
    "    #Sort the DataFrame in order decreasing feature importance\n",
    "    fi_df.sort_values(by=['feature_importance'], ascending=False,inplace=True)\n",
    "\n",
    "    #Define size of bar plot\n",
    "    plt.figure(figsize=(10,8))\n",
    "    #Plot Searborn bar chart\n",
    "    sns.barplot(x=fi_df['feature_importance'], y=fi_df['feature_names'])\n",
    "    plt.ylim(0, 50)\n",
    "    #Add chart labels\n",
    "    plt.title(model_type + 'FEATURE IMPORTANCE')\n",
    "    plt.xlabel('FEATURE IMPORTANCE')\n",
    "    plt.ylabel('FEATURE NAMES')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_feature_importance(rf_clf.feature_importances_,x_train.columns,'RANDOM FOREST') \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 800x560 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x1 = reduced_dataset.loc[reduced_dataset.isMalware==1, 'num_of_exts']\n",
    "x2 = reduced_dataset.loc[reduced_dataset.isMalware==0, 'num_of_exts']\n",
    "\n",
    "kwargs = dict(hist_kws={'alpha':.6}, kde_kws={'linewidth':2})\n",
    "\n",
    "plt.figure(figsize=(10,7), dpi= 80)\n",
    "sns.distplot(x1, color=\"royalblue\", label=\"Malware\", **kwargs, kde=False)\n",
    "sns.distplot(x2, color=\"orange\", label=\"Benign\", **kwargs, kde=False)\n",
    "\n",
    "plt.legend();\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqYAAAHZCAYAAABU5/kWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAMTQAADE0B0s6tTgAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3dfbCeZX0n8O8vJE22RFtkIYghBEugoishBhVFRQQr3Ra3QNnRAXlpd+modaZU6hSY1ul0R7SWdmx1Gju6UIV2R20p2K62iAgsIFoJvtASqhxDgpUapRB5Dbn2j/OceAgJnpfn5Fw5+Xxm7uE81+++7+e6Lw/HL9f9Vq21AADAbJs32x0AAIBEMAUAoBOCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0IUJBdOq+kBVjVRVq6oX7aD+u9vXqmpFVd1cVeuq6raqOmK6NQAA5q75E1zvk0nel+Sm7QtVtSrJy5Os3660JsmHW2uXVdVpST6S5Jhp1p7RwoUL23777TfBQwIAYFfbuHHj4621hTuq1WTe/FRVI0l+obX29cHnhUmuT/LmJJ8fq1XV/knWJfnPrbUtVVVJvpPRAPvwVGqttZEf17+lS5e2DRs2TPh4AADYtapqY2tt6Y5q073G9PeSfLy1ds927Qclua+1tiVJ2mj6XZ9k2TRqAADMYVMOplV1TJKjk3xoJ6tsPxVbQ6ht34fzq2rD2LJ58+Zn6jIAAB2bzozpa5L8bJJ7Bqf4lyb5bFWdlOTeJEuran6SDE7JH5TR2c+p1p6mtXZpa23p2LJ48eJpHA4AALNpojc/PU1r7ZIkl4x93sH1p7cnOSPJZUlOTTIydp3oVGsAADOhtbZtYXqqKvPmTW3uc0I3P1XVB5O8MckBSb6XZHNr7dDt1hnJU4Pp4RkNl/smeTDJWa21b0yn9uO4+QkAmIytW7fm/vvvzwMPPCCUDtGCBQuybNmy/MRP/MTTas9089Ok7srvnWAKAEzGPffck3nz5mXJkiVZsGDBbHdnTmitZdOmTXnooYdy6KGHPq3+TMF0yqfyAQB2Z1u3bs2jjz6aFStWZP58kWiY9t1333z/+9/P1q1bJ3Va3ytJAYA90thZ49F7rRmmsTGd7Jl5/3kAADDOpVd+f0b2e/6bnzOh9ZYvX56HH344Gzdu3HZ5wXXXXZfXve51+c3f/M28//3v3+m2Z599dlavXp23v/3tQ+nzrmbGFACgM8uWLcvVV1+97fNHP/rRrF69etb6s2XLll3yPYIpAEBnzj333Hz0ox9NkvzHf/xHbr311rzhDW9Iknzta1/Lq171qqxatSpHHHFE3vOe9+xwH8973vNy3333JUlOOeWUvPKVr0ySPPLII3nOc56Txx577Bn3dfbZZ+cd73hH3vCGN+TII49MknzpS1/K8ccfn9WrV2fVqlX51Kc+NdTjdiofAKAzr371q/Mnf/In2bhxY6655pr88i//cvbaa68ko6f6r7322ixcuDCPPPJIXvGKV+TEE0982ozq8ccfn2uvvTZnnHFGvva1r2XhwoV56KGHcvPNN+foo4/OwoULf+y+brrpptxwww1ZvHhxHnjggZx33nn5u7/7uzz3uc/N9773vbzkJS/JK1/5yhxwwAFDOW7BFACgQ2eeeWYuv/zyXHXVVbniiityxRVXJBmd8XzrW9+atWvXZt68ebn33nuzdu3apwXTE044Iddee21e8IIX5KijjsqSJUvyhS98IV/4whdywgknTGhfp59+esberHnzzTfnW9/6Vk466aRt39Fay1133SWYAgDMZWeffXZWrVqVww47LCtWrNjWfuGFF2bJkiW5/fbbM3/+/Jxyyil59NFHn7b9iSeemAsvvDAveMELcsIJJ2TJkiX53Oc+l+uvv37bZQI/bl/jX/feWsuLX/zi3HDDDTN2zK4xBQDo0IEHHpj3vOc9ee973/uU9h/84AdZunRp5s+fn7vuuiv/+I//uNPtn/3sZ2fNmjU54YQT8trXvjZXX311Nm7cmJUrV05qX0nyile8InfffXeuu+66bW1r167N448/PoSjHWXGFACgU+ecc87T2i6++OKceeaZueKKK7J8+fIcf/zxO93+xBNPzKc//ek8//nPT5IsWbIkq1ev3vac0cnsa5999sk111yTCy64IL/xG7+RJ554IsuWLctVV101zaP8Ea8kBQD2SE8++WTWrVuXww47bNuNRQzHM43tM72S1Kl8AAC64FQ+AOxmJvNmoom+bQh6YMYUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogrvyAQDGu+28mdnvS9dMaLXly5dn0aJFWbRoUR577LEcddRR+fM///PsvffeU/raq6++OjfeeGP+4A/+YErb70pmTAEAOvPJT34ya9euzZ133pkHH3wwl1122ZT3dfLJJ+8WoTQRTAEAuvXYY4/lhz/8YfbZZ58kyfvf//689KUvzapVq/LzP//zuffee5Mk7373u/PmN785v/iLv5gjjjgixx9/fL7//dHn3V522WU57bTTtu3zoosuyqGHHpqXvexlueCCC7J69eokyfXXX5+VK1fmrW99a4488si88IUvzJe//OVderyCKQBAZ0477bSsXLkyS5YsSVXl9NNPz5VXXpl169bllltuyVe+8pW86U1vytvf/vZt23zxi1/M5ZdfnjvvvDP7779/1qx5+qUD11xzTT796U/njjvuyC233JJvfvObT6l/4xvfyLnnnps77rgjv/7rv56LLrpoxo91PMEUAKAzY6fyN23alEMOOSTvete7ctVVV+Xaa6/NS17ykqxcuTLve9/78u1vf3vbNieddFKe85zRN30dc8wxTwudSfL5z38+p59+evbee+/MmzcvZ5111lPqhx9++LYZ1J3tYya5+QkAoFPz58/PqaeemgsuuCBHHHFELr744px77rk7XHfRokXbft5rr72yZcuWp63TWktV7fT7JrKPmWTGFACgY9ddd10OP/zwnHzyyfnQhz607drRJ554Irfffvuk9vXa1742n/jEJ/Lwww9n69at+djHPjYTXZ4yM6YAAJ057bTTsmjRojzxxBNZvnx5/uzP/iwHHXRQNm3alOOOOy5VlS1btuRXfuVXctRRR014vyeffHJuvvnmHHnkkTnwwAPz8pe/PD/4wQ9m8Egmp1prs92HoVm6dGnbsGHDbHcDAGbUpVd+f8Lrnv/m58xgT3ZvTz75ZNatW5fDDjsse+2112x3Z5d56KGH8qxnPStbt27Nr/7qr+bAAw/M7//+7w/1O55pbKtqY2tt6Y62M2MKALAHectb3pKRkZE88sgjWbVqVX7rt35rtru0jWAKALAH+Zu/+ZvZ7sJOufkJAIAuCKYAwB5p7LFJc+l+m16MjekzPZpqR5zKBwD2SPPmzcuiRYuycePGLFmyJAsWLJjtLs0JrbVs2rQpCxYsyLx5k5sDFUwBgD3WwQcfnPvvvz8jIyNmTodowYIFWbZs2aS3E0wBgD3WvHnzcsABB2TJkiVprQmnQ1BVk54pHSOYAgB7vKqa9PWQDJ+bnwAA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXZhQMK2qD1TVSFW1qnrRoG1RVV1VVeuqam1Vfaaqlo/bZv9B291V9fWqOna6NQAA5q6Jzph+MsmxSb69XfuHkxzeWluZ5NODz2MuSXJra21FknOSXFFV86dZAwBgjppQMG2t3dBa27Bd26Ottb9vrbVB061Jnj9uldOTfHCw7peSfDej4XY6NQAA5qhhXmP6jiTXJElV7ZtkXmvt38fVR5Ism2ptiP0EAKBDQwmmVXVhkhVJLhrX3LZfbQi17b/3/KraMLZs3rx5ol0GAKAz0w6mVfXOJKckOam19nCStNY2DWr7jVv14CTrp1rb0Xe31i5trS0dWxYvXjzdwwEAYJZMK5hW1flJ3pTkxNbaA9uVP5HkbYP1jk5yQJKbplkDAGCOmtDd7lX1wSRvzGhIvLaqNic5LskfJvlWks9XVZI81lp72WCzdyX5WFXdneTxJGe21rZMswYAwBw1oWDaWntbBrOY29np9Z+tte8mef0wawAAzF3e/AQAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQhQkF06r6QFWNVFWrqheNa19RVTdX1bqquq2qjpjJGgAAc9dEZ0w/meTYJN/ern1Nkg+31g5L8r4kH5nhGgAAc9SEgmlr7YbW2obxbVW1f5JVST4+aPpUkkOqavlM1KZycAAA7D7mT2Pbg5Lc11rbkiSttVZV65MsS/LDGaiNTKOvAAB0bro3P7XtPtcM155aqDq/qjaMLZs3b37GzgIA0K/pBNN7kyytqvlJUlWV0VnU9TNUe5rW2qWttaVjy+LFi6dxOAAAzKYpB9PW2v1Jbk9yxqDp1CQjrbWRmahNtZ8AAOweJnSNaVV9MMkbkxyQ5Nqq2txaOzTJeUkuq6oLkzyY5Kxxm81EDQCAOWpCwbS19rYkb9tB+11JjtnJNkOvAQAwd3nzEwAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXhhJMq+rnquqfqur2qvp6VZ01aN+/qj5TVXcP2o8dt82UagAAzE3TDqZVVUmuTHJOa+2oJL+QZE1VPSvJJUluba2tSHJOkiuqav5g06nWAACYg4YZ9n568M9nJ9mU5LEkpyc5JElaa1+qqu8mOTbJ9dOoAQAwB007mLbWWlWdnuSvq+qHSfZJckqSZyWZ11r793GrjyRZVlX7TqU23b4CANCvYZzKn5/kt5O8sbV2cJLXJbl8UG7brz7u56nWxn/3+VW1YWzZvHnz5DoPAEA3hnHz08okB7bW/l8yeuo9yX1JXpwkVbXfuHUPTrK+tbZpKrXtv7i1dmlrbenYsnjx4iEcDgAAs2EYwfTeJEur6vAkqapDk/xMknVJPpHkbYP2o5MckOSmwXZTrQEAMAcN4xrT71bVeUk+WVVbM3ra/a2ttY1V9a4kH6uqu5M8nuTM1tqWwaZTrQEAMAcN5a781tpfJvnLHbR/N8nrd7LNlGoAAMxN3vwEAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0IWhBNOqWlhVf1pVd1fVN6rq44P2FVV1c1Wtq6rbquqIcdtMqQYAwNw0rBnTS5JsTXJYa+2FSS4YtK9J8uHW2mFJ3pfkI+O2mWoNAIA5aNrBtKr2TnJOkgtbay1JWmvfqar9k6xK8vHBqp9KckhVLZ9qbbp9BQCgX8OYMf2ZJJuSXFxVX66qG6vqdUkOSnJfa21LkgxC6/oky6ZRAwBgjhpGMF2Q5PlJ7mytrU7y9iR/lWR+krbdujXu56nWftRYdX5VbRhbNm/ePOnOAwDQh2EE029n9PrSK5KktXZHknuSHJxkaVXNT5KqqozOhq5Pcu8Ua0/RWru0tbZ0bFm8ePEQDgcAgNkw7WDaWvteks8l+bkkqaqDkxyS5MYktyc5Y7DqqUlGWmsjrbX7p1Kbbl8BAOjX/CHt59eSfLSq3pvkyST/c3AD1HlJLquqC5M8mOSscdtMtQYAwBw0lGDaWvtWkuN20H5XkmN2ss2UagAAzE3e/AQAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0Yf5sdwCAPdht501u/ZeumZl+AF0wYwoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHTBXfkADNdk77QHGDBjCgBAFwRTAAC64FQ+AEzm8gMP+YcZY8YUAIAumDEF4MdzQxOwC5gxBQCgC4IpAABdGGowrarfrapWVS8afF5RVTdX1bqquq2qjhi37pRqAADMTUO7xrSqViV5eZL145rXJPlwa+2yqjotyUeSHDPNGgA74jpQYDc3lBnTqlqY5INJ3pqkDdr2T7IqyccHq30qySFVtXyqtWH0FQCAPg3rVP7vJfl4a+2ecW0HJbmvtbYlSVprLaOzqcumUQMAYI6adjCtqmOSHJ3kQzsot+1XH0Jt/HefX1UbxpbNmzdPpMsAAHRoGDOmr0nys0nuqaqRJEuTfDbJi5Israr5SVJVldHZ0PVJ7p1i7Slaa5e21paOLYsXLx7C4QAAMBumffNTa+2SJJeMfR6E019orX29qs5JckaSy5KcmmSktTYyWO/2qdQAYELcDAa7nZl+89N5SS6rqguTPJjkrCHUAACYg4YeTFtry8f9fFd28pinqdYAYFZNdib2pWtmph8wB3nzEwAAXRBMAQDogmAKAEAXZvrmJwAYHnfaw5wmmAL0TBAD9iBO5QMA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuCKYAAHRBMAUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOiCYAoAQBcEUwAAuiCYAgDQBcEUAIAuzJ/tDgDAnHbbeRNf96VrZq4fsBswYwoAQBcEUwAAuiCYAgDQhWkH06paVFVXVdW6qlpbVZ+pquWD2v6Dz3dX1der6thx202pBgDA3DSsGdMPJzm8tbYyyacHn5PkkiS3ttZWJDknyRVVNX+aNQAA5qBpB9PW2qOttb9vrbVB061Jnj/4+fQkHxys96Uk301y7DRrAADMQTNxjek7klxTVfsmmdda+/dxtZEky6Zam4G+AgDQiaEG06q6MMmKJBcNmtr2q4z7eaq18d93flVtGFs2b9482S4DANCJoQXTqnpnklOSnNRae7i1tmnQvt+41Q5Osn6qte2/s7V2aWtt6diyePHiYR0OAAC72FCCaVWdn+RNSU5srT0wrvSJJG8brHN0kgOS3DTNGgAAc9C073SvqqVJ/jDJt5J8vqqS5LHW2suSvCvJx6rq7iSPJzmztbZlsOlUawAAzEHTDqattQ3ZyTWgrbXvJnn9MGsAAMxN3vwEAEAXBFMAALogmAIA0AXBFACALnj/PMCudtt5s90DgC6ZMQUAoAuCKQAAXRBMAQDogmAKAEAXBFMAALogmAIA0AXBFACALniOKQCw+5vM84Ffumbm+sG0mDEFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFwRQAgC54jikA7Ak855PdgBlTAAC6IJgCANAFp/IBYDd0Qs6f2Iq3LZzZjsAQmTEFAKALgikAAF1wKh8AYC6ZzBMYkq6ewmDGFACALpgxBQCeyjNPmSWCKQD0YoKB8IQ8NsMdgdnhVD4AAF0wYwoATJ3T/gyRYAoA7Bq78d3i7BpO5QMA0AXBFACALjiVDwCwMy4/2KUEUwBgzzLZsMkuI5gCAH0SIPc4rjEFAKALgikAAF0QTAEA6IJgCgBAFwRTAAC6IJgCANAFj4sCgDnsq//62ITXffGhC2ewJ3uIyTziysP4n0YwBQDo3R7yTFen8gEA6IJgCgBAFwRTAAC64BpTACDJ5G6UStwsxfCZMQUAoAvdBtOqWlFVN1fVuqq6raqOmO0+AQAwc3o+lb8myYdba5dV1WlJPpLkmFnuEwAwsDs+I3V37POepMtgWlX7J1mV5PWDpk8l+dOqWt5aG5m1jgEAUzLZ61eZmt09eHcZTJMclOS+1tqWJGmttapan2RZkpHZ7Nh4X73y7Amv++JDF3rDA3uuyT4Y2r8rwG5mJgPhpVd+Pydkzwj2vQbTJGnbfa7tV6iq85OcP67pyar6tyF9/+Ikm4e0r4EPD3d3u68ZGFvGmQPj2+2/K3NgbLtmfGeOsZ050xjbXv7W7fJ+7LezQrW2ff6bfYNT+Xcn2be1tqWqKsl3krx8V53Kr6oNrbWlu+K79jTGdmYZ35ljbGeW8Z05xnbmGNvh6vKu/Nba/UluT3LGoOnUJCOuLwUAmLt6PpV/XpLLqurCJA8mOWuW+wMAwAzqNpi21u7K7D4e6tJZ/O65ztjOLOM7c4ztzDK+M8fYzhxjO0RdXmMKAMCep8trTAEA2PMIpgAAdEEw3U5Vraiqm6tqXVXdVlVHzHafelNVH6iqkapqVfWice07HbuZqM1FVbWoqq4aHO/aqvpMVS0f1PYffL67qr5eVceO227otbmqqv6hqr46GN8bq2rloN3v75BU1e+O//tgbKdv8Df3Xwa/t2ur6r8P2o3tEFTVwqr608Hfwm9U1ccH7cZ3V2utWcYtSa5Lcvbg59OS3DLbfeptSfLqJEsz+hauF01k7GaiNheXJIuS/Hx+dP3325P8w+DnjyZ59+Dno5N8O8n8marN1SXJT4/7+b8l+cp0fg/9/j5tfFcl+b+D36UXGduhjetT/t7O5PjtaWM7OM4/SvKBcX97n2t8Z+l/i9nuQE9Lkv2TPDDu/7Qryb8lWT7bfetxGf+H8pnGbiZqs33su3CMVyf518HPm5PsN652W5LjZqq2JywZfQzdl/3+Dm08Fya5JckhY38fjO3QxnYk2wVTYzu0sd17cMyLje/sL90+LmqWHJTkvtbaliRprbWqWp9kWUb/KLBzzzR2P5yB2sguPbrZ844k11TVvknmtdb+fVxtJMmymagN/Sg6U1V/keS1g49viN/fYfm9JB9vrd1Tte0t0sZ2eK6oqnlJvpjkt2Nsh+VnkmxKcnFVnZDkkSTvzmiANL67mGtMn27752fVDtdiR55p7GaiNqfV6MslViS5aNBkfIektfaW1tpBSS5O8gdjzdutZnwnoaqOyejlIB/aQdnYTt+rW2tHZvRSiU1JLh+0G9vpW5Dk+UnubK2tzuglVH+V0We9G99dTDB9qnuTLK2q+UlSo//Jf1CS9bPaq93DM43dTNTmtKp6Z5JTkpzUWnu4tbZp0L7fuNUOTrJ+JmrDPp5etdYuz+jM6Yb4/Z2u1yT52ST3VNVIRq9D/2xGT+cb22lqra0f/POJJH+c5FXxd3dYvp1ka5IrkqS1dkeSezL699D47mKC6TittfuT3J7kjEHTqUlGWmsjs9ap3cQzjd1M1Gb8gGZRVZ2f5E1JTmytPTCu9Ikkbxusc3SSA5LcNIO1Oaeqnl1VB477/EsZnX3y+ztNrbVLWmsHttaWt9aWZzTs/9wg/Bvbaaiqvavqp8c1vSnJ7f7uDkdr7XtJPpfk55Kkqg7O6HXSN8b47nqtgwtde1qSHJ7Ri/fXZfSmiBfOdp96W5J8MKP/p7Mloxdtj92cs9Oxm4naXFwyOsvUknwzydrB8sVBbUmSf0hyd5JvJHnNuO2GXpuLS0ZnJm5L8rUkdyS5NslKv78zMtYj+dHNkcZ2emP5/IyGma8Ofnf/NoObZYztUMf4+sH4rk3yS8Z3dhavJAUAoAtO5QMA0AXBFACALgimAAB0QTAFAKALgikAAF0QTAEA6IJgCjAkVXVKVf1TVa2tqn+uqs8N3m0+3f2+u6ruH+z3zqr6P1W1zxT2c1xVvX66/QGYKYIpwBBU1QFJ/izJKa21la21FyS5INu9F3vsVYRT8BettZVJ/kuSvZJcPMn+zU9yXBLBFOiWYAowHM/N6NvQNo01tNa+0lprVTVSVRdV1eeTXJ4kVXXOYAb0jqr6clUtn8iXtNaeTHJdRt8ek6paXVW3VNVXq+q2qnrloH15VX2vqn6nqm5M8utJfi3JWwbf+zvDO3SA4Zjqf7kD8FR3ZPQ1g+ur6gtJbk5yZWtt46C+LMnxg6B6XJKLkryqtfadqvrJiX5JVf2nJG9McmtV/USSvwS80QYAAAF/SURBVE7yP1prn62qY5N8sqoOHay+b0ZfGfx7g21/Ksni1to7p320ADPAjCnAELTWtrbWTk3yiiSfSfLKJN8YFxL/d/vRO6D/a0ZPzX9nsO3DrbWHf8xXvKWqbs9o4P3nJJdkdNb08dbaZwf7uSnJ/UlePNjm0SR/OZQDBNgFzJgCDFFr7V+S/EuSNVX1mSQnD0qbp7nrv9h+prOqKttdwzrWjcE/fzguDAN0z4wpwBBU1fPGru8cfN4nySFJvrmD1a/J6AzoAYN1f3Iyp/PH+ZckC6vq+MF+XpFk/yRf28n6Dyb5qSl8D8AuIZgCDMf8JL9TVeuqam2SG5Nc3lr72+1XbK3dkOT3k/xDVd2R5AtJ9pvsF7bWHk9yapL/VVVfTfLHSX65tfbDnWzyN0lWu/kJ6FU5ywMAQA/MmAIA0AU3PwF0oKpWJrlsB6XLW2t/tIu7AzArnMoHAKALTuUDANAFwRQAgC4IpgAAdEEwBQCgC4IpAABdEEwBAOjC/weuHGL76OME6QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 800x560 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x1 = reduced_dataset.loc[reduced_dataset.isMalware==1, 'Src_Port']\n",
    "x2 = reduced_dataset.loc[reduced_dataset.isMalware==0, 'Src_Port']\n",
    "\n",
    "kwargs = dict(hist_kws={'alpha':.6}, kde_kws={'linewidth':2})\n",
    "\n",
    "plt.figure(figsize=(10,7), dpi= 80)\n",
    "sns.distplot(x1, color=\"royalblue\", label=\"Malware\", **kwargs, kde=False)\n",
    "sns.distplot(x2, color=\"orange\", label=\"Benign\", **kwargs, kde=False)\n",
    "\n",
    "plt.legend();\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Classifier Accuracy score:  0.9951428571428571\n"
     ]
    }
   ],
   "source": [
    "#Saving the random forest and logistics regression models\n",
    "joblib.dump(rf_clf, 'random_forest_model.pkl')\n",
    "joblib.dump(lm, 'logistics_regression_classifier.pkl')\n",
    "\n",
    "# Load the model from the file\n",
    "rf_from_joblib = joblib.load('random_forest_model.pkl')\n",
    "lm_from_joblib = joblib.load('logistics_regression_classifier.pkl') \n",
    "\n",
    "# Use the loaded model to make predictions\n",
    "results = rf_from_joblib.predict(x_test)\n",
    "results = lm_from_joblib.predict(x_test)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(y_test, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the model from the file\n",
    "rf_from_joblib = joblib.load('random_forest_model.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Classifier Accuracy score:  0.8588692536588328\n"
     ]
    }
   ],
   "source": [
    "#Cipher and extension stunded dataset tests\n",
    "cs_ex_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "cs_ex_stunted_dataset = cs_ex_stunted_dataset[cs_ex_stunted_dataset.duplicated(keep=False) == False]\n",
    "cs_ex_test_x, cs_ex_test_y = cs_ex_stunted_dataset.drop(['isMalware'], axis=1), cs_ex_stunted_dataset['isMalware']\n",
    "results = rf_from_joblib.predict(cs_ex_test_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(cs_ex_test_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Classifier Accuracy score:  0.5675593611465355\n"
     ]
    }
   ],
   "source": [
    "#Cipher, extension, and source port stunted results\n",
    "cs_ex_sp_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "cs_ex_sp_stunted_dataset = cs_ex_sp_stunted_dataset[cs_ex_sp_stunted_dataset.duplicated(keep=False) == False]\n",
    "cs_ex_sp_test_x, cs_ex_sp_test_y = cs_ex_sp_stunted_dataset.drop(['isMalware'], axis=1), cs_ex_sp_stunted_dataset['isMalware']\n",
    "results = rf_from_joblib.predict(cs_ex_sp_test_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(cs_ex_sp_test_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Forest Classifier Accuracy score:  0.6059998785449687\n"
     ]
    }
   ],
   "source": [
    "#Cipher, extension, source port, and packet out stunted results\n",
    "cs_ex_sp_po_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "cs_ex_sp_po_stunted_dataset = cs_ex_sp_po_stunted_dataset[cs_ex_sp_po_stunted_dataset.duplicated(keep=False) == False]\n",
    "cs_ex_sp_po_test_x, cs_ex_sp_po_test_y = cs_ex_sp_po_stunted_dataset.drop(['isMalware'], axis=1), cs_ex_sp_po_stunted_dataset['isMalware']\n",
    "results = rf_from_joblib.predict(cs_ex_sp_po_test_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(cs_ex_sp_po_test_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Training new models with original and cipher stunted data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1    82335\n",
      "0    62604\n",
      "Name: isMalware, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Combining all datasets together\n",
    "new_all_flows = pd.concat([malicious_dataset, benign_dataset, cipher_stunted_dataset, cs_ex_stunted_dataset, cs_ex_sp_stunted_dataset, cs_ex_sp_po_stunted_dataset])\n",
    "\n",
    "# Examining combined dataset\n",
    "print(new_all_flows['isMalware'].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Isolating independent and dependent variables for new combined dataset\n",
    "new_all_flows_y = new_all_flows['isMalware']\n",
    "new_all_flows_x = new_all_flows.drop(['isMalware'], axis=1);\n",
    "\n",
    "# Splitting datasets into training and test data\n",
    "combined_x_train, combined_x_test, combined_y_train, combined_y_test = train_test_split(new_all_flows_x, new_all_flows_y, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[12541     0]\n",
      " [   10 16437]]\n",
      "Random Forest Classifier Accuracy score:  0.9996550296674486\n"
     ]
    }
   ],
   "source": [
    "#Training new RF classifier\n",
    "combined_rf_clf = RandomForestClassifier(max_depth=100)\n",
    "combined_rf_clf.fit(combined_x_train, combined_y_train) \n",
    "combined_rf_prediction = combined_rf_clf.predict(combined_x_test)\n",
    "combined_conf_m = confusion_matrix(combined_y_test, combined_rf_prediction)  \n",
    "print(combined_conf_m)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(combined_y_test, combined_rf_prediction))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Testing combined Logistics Regression classifier on datasets\n",
    "combined_lm = LogisticRegression(max_iter=100000)\n",
    "combined_lm.fit(combined_x_train,combined_y_train)\n",
    "combined_lm_predictions = combined_lm.predict(combined_x_test)\n",
    "combined_rms = sqrt(mean_squared_error(combined_y_test, combined_lm_predictions))\n",
    "print('Mean Absolute Error: ', mean_absolute_error(combined_y_test, combined_lm_predictions))\n",
    "print('Root Mean Squared Error: ',combined_rms)\n",
    "print('R Squared Score: ', r2_score(combined_y_test, combined_lm_predictions))\n",
    "print('Accuracy score: ', accuracy_score(combined_lm_predictions, combined_y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "combined_rf_prediction_2 = combined_rf_clf.predict(sample_benign_flows_x)\n",
    "\n",
    "combined_lm_predictions_2 = combined_lm.predict(sample_benign_flows_x)\n",
    "\n",
    "# Testing on unseen sample benign dataset for combined rf classifier\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(sample_benign_flows_y, combined_rf_prediction_2))\n",
    "\n",
    "# Testing on unseen sample benign dataset for logistics regression classifier\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(sample_benign_flows_y, combined_lm_predictions_2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Cipher stunted results\n",
    "new_cs_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "new_cs_stunted_dataset = new_cs_stunted_dataset[new_cs_stunted_dataset.duplicated(keep=False) == False]\n",
    "new_cs_stunted_dataset_x, new_cs_stunted_dataset_y = new_cs_stunted_dataset.drop(['isMalware'], axis=1), new_cs_stunted_dataset['isMalware']\n",
    "results = combined_rf_clf.predict(new_cs_stunted_dataset_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(new_cs_stunted_dataset_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Cipher and extension stunted results\n",
    "new_cs_ex_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "new_cs_ex_stunted_dataset = new_cs_ex_stunted_dataset[new_cs_ex_stunted_dataset.duplicated(keep=False) == False]\n",
    "new_cs_ex_stunted_dataset_x, new_cs_ex_stunted_dataset_y = new_cs_ex_stunted_dataset.drop(['isMalware'], axis=1), new_cs_ex_stunted_dataset['isMalware']\n",
    "results = combined_rf_clf.predict(new_cs_ex_stunted_dataset_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(new_cs_ex_stunted_dataset_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Cipher, extension, and source port stunted results\n",
    "new_cs_ex_sp_stunted_malicious_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "new_cs_ex_sp_stunted_malicious_dataset = new_cs_ex_sp_stunted_malicious_dataset[new_cs_ex_sp_stunted_malicious_dataset.duplicated(keep=False) == False]\n",
    "new_cs_ex_sp_stunted_malicious_dataset_x, new_cs_ex_sp_stunted_malicious_dataset_y = new_cs_ex_sp_stunted_malicious_dataset.drop(['isMalware'], axis=1), new_cs_ex_sp_stunted_dataset['isMalware']\n",
    "results = combined_rf_clf.predict(new_cs_ex_sp_stunted_malicious_dataset_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(new_cs_ex_sp_stunted_malicious_dataset_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Cipher, extension, source port, and packets out stunted results\n",
    "new_cs_ex_sp_po_stunted_dataset = pd.read_csv('Insert cipher stunted file name')\n",
    "new_cs_ex_sp_po_stunted_dataset = new_cs_ex_sp_po_stunted_dataset[new_cs_ex_sp_po_stunted_dataset.duplicated(keep=False) == False]\n",
    "new_cs_ex_sp_po_stunted_dataset_x, new_cs_ex_sp_po_stunted_dataset_y = new_cs_ex_sp_po_stunted_dataset.drop(['isMalware'], axis=1), new_cs_ex_sp_po_stunted_dataset['isMalware']\n",
    "results = combined_rf_clf.predict(new_cs_ex_sp_po_stunted_dataset_x)\n",
    "print('Random Forest Classifier Accuracy score: ', accuracy_score(new_cs_ex_sp_po_stunted_dataset_y, results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "lm_predictions_cs_ex = combined_lm.predict(new_cs_ex_stunted_dataset_x)\n",
    "print('Accuracy score: ', accuracy_score(lm_predictions_cs_ex, new_cs_ex_stunted_dataset_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "lm_predictions_cs_ex_sp = combined_lm.predict(new_cs_ex_sp_stunted_malicious_dataset_x)\n",
    "print('Accuracy score: ', accuracy_score(lm_predictions_cs_ex_sp, new_cs_ex_sp_stunted_malicious_dataset_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "lm_predictions_cs_ex_sp_po = combined_lm.predict(new_cs_ex_sp_po_stunted_dataset_x)\n",
    "print('Accuracy score: ', accuracy_score(lm_predictions_cs_ex_sp_po, new_cs_ex_sp_po_stunted_dataset_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "joblib.dump(combined_rf_clf, 'new_random_forest_model.pkl')\n",
    "joblib.dump(combined_lm, 'new_logistics_regression_classifier.pkl')"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}

一键复制编辑原始数据按行查看历史

提交于 2020-08-15 00:13 . Rename file

马建仓 AI 助手

尝试更多

代码解读

代码找茬

代码优化

https://gitee.com/wu_zhengqi/TLS-Malware-Detection-with-Machine-Learning.git

[email protected]:wu_zhengqi/TLS-Malware-Detection-with-Machine-Learning.git

wu_zhengqi

TLS-Malware-Detection-with-Machine-Learning

master

吴政琪/TLS-Malware-Detection-with-Machine-Learning

简介

发行版

贡献者

近期动态

吴政琪/TLS-Malware-Detection-with-Machine-Learning .gitee-modal { width: 500px !important; }

简介

发行版

贡献者

近期动态

搜索帮助

吴政琪/TLS-Malware-Detection-with-Machine-Learning