Supervised_TFIDF_100k.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "executionInfo": {
     "elapsed": 6763,
     "status": "ok",
     "timestamp": 1622242183534,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "6bPVlN6RcK52"
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "df_original = pd.read_csv('sample_data/WikiLarge_Train.csv')\n",
    "df_original['re_tokened'] = pickle.load(open('sample_data/re_tokenized_lemma.pkl', 'rb'))\n",
    "df_original['re_tokened2'] = df_original['re_tokened'].apply(lambda x: ' '.join(x))\n",
    "\n",
    "df = df_original.sample(100000, random_state=42)\n",
    "Train_X, Test_X, Train_Y, Test_Y = train_test_split(df['re_tokened2'],df['label'],test_size=0.2,random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "executionInfo": {
     "elapsed": 1849,
     "status": "ok",
     "timestamp": 1622242186557,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "arPhrMawdfsh"
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "\n",
    "Tfidf_vect = TfidfVectorizer(max_features=10000, lowercase=True, stop_words='english')\n",
    "\n",
    "Train_X_Tfidf = Tfidf_vect.fit_transform(Train_X)\n",
    "Test_X_Tfidf = Tfidf_vect.transform(Test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 203,
     "status": "ok",
     "timestamp": 1622242190682,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "PO2GUAKwZBx8",
    "outputId": "beb2dfc2-60d0-4726-86a5-76a712b4e23c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((80000, 10000), (20000, 10000))"
      ]
     },
     "execution_count": 3,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Train_X_Tfidf.shape, Test_X_Tfidf.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HHDuOVX-c5eq"
   },
   "source": [
    "# Skipped with SVM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Dujx5ziYeQu0"
   },
   "source": [
    "# MLP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 3475546,
     "status": "ok",
     "timestamp": 1622103988744,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "DFBk1wULeRvI",
    "outputId": "c71e6c39-83a3-471a-9a89-64ca564cf0e1"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(63.77, 63.77724455108977)"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.neural_network import MLPClassifier\n",
    "\n",
    "mlp = MLPClassifier(hidden_layer_sizes=(150,), random_state=42, max_iter=300)\n",
    "mlp.fit(Train_X_Tfidf, Train_Y)\n",
    "mlp_preds = mlp.predict(Test_X_Tfidf)\n",
    "mlp_acc = accuracy_score(Test_Y, mlp_preds)*100\n",
    "mlp_f1 = f1_score(Test_Y, mlp_preds)*100\n",
    "mlp_acc, mlp_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "PK9wqzrSugW_"
   },
   "outputs": [],
   "source": [
    "pickle.dump(mlp, open('drive/MyDrive/Milestone2/trained_models/mlp_tfidf.pkl', 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "x7u3SEycezl1"
   },
   "outputs": [],
   "source": [
    "reloaded_mlp = pickle.load(open('drive/MyDrive/Milestone2/trained_models/mlp_tfidf.pkl', 'rb'))\n",
    "mlp_preds = reloaded_mlp.predict(Test_X_Tfidf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 145,
     "status": "ok",
     "timestamp": 1622130906856,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "h1iSZ3cyt3BF",
    "outputId": "c271ba67-0454-4943-9ea7-8a3657c2fa7f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(63.77, 63.77724455108977)"
      ]
     },
     "execution_count": 9,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mlp_acc = accuracy_score(Test_Y, mlp_preds)*100\n",
    "mlp_f1 = f1_score(Test_Y, mlp_preds)*100\n",
    "mlp_acc, mlp_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 410
    },
    "executionInfo": {
     "elapsed": 875,
     "status": "ok",
     "timestamp": 1622133632221,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "_Dc20l7vt3Dr",
    "outputId": "c02ae8b1-5999-43d5-cbc0-3be4fd3b3cd3"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score, roc_curve, auc\n",
    "\n",
    "y_score = reloaded_mlp.predict_proba(Test_X_Tfidf)\n",
    "mlp_fpr, mlp_tpr, thresholds = roc_curve(Test_Y, y_score[:,1], pos_label=1)\n",
    "area = auc(mlp_fpr, mlp_tpr)\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_theme()\n",
    "\n",
    "plt.figure(figsize=(8,6))\n",
    "lw = 2\n",
    "plt.plot(mlp_fpr, mlp_tpr, color='darkorange',\n",
    "         lw=lw, label='ROC curve (area = %0.2f)' % area)\n",
    "plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('ROC score MLP with Tf-Idf')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.savefig('ROC_MLP_tfidf.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "eTYF9RL2t3Gl"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "uMTjj_Mit3J8"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8IRNq-FNd7F_"
   },
   "source": [
    "# Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 69668,
     "status": "ok",
     "timestamp": 1622132600885,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "kGcIwlDmd6cs",
    "outputId": "b0b2c78f-fca3-4eb8-c881-a3302cea4673"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.6498, 0.6207904710341093)"
      ]
     },
     "execution_count": 12,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "rf = RandomForestClassifier(bootstrap= True,\n",
    " max_depth=70,\n",
    " max_features='auto',\n",
    " min_samples_leaf=4,\n",
    " min_samples_split=10,\n",
    " n_estimators=800)\n",
    "\n",
    "rf.fit(Train_X_Tfidf, Train_Y)\n",
    "rf_preds = rf.predict(Test_X_Tfidf)\n",
    "rf_acc = accuracy_score(Test_Y, rf_preds)\n",
    "rf_f1 = f1_score(Test_Y, rf_preds)\n",
    "rf_acc, rf_f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "executionInfo": {
     "elapsed": 3442,
     "status": "ok",
     "timestamp": 1622242222734,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "8fazq4CACVrT"
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "pickle.dump(rf, open('drive/MyDrive/Milestone2/trained_models/rf_tfidf_10w.pkl', 'wb'))\n",
    "reloaded_rf = pickle.load(open('drive/MyDrive/Milestone2/trained_models/rf_tfidf_10w.pkl', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 410
    },
    "executionInfo": {
     "elapsed": 4816,
     "status": "ok",
     "timestamp": 1622242230609,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "oOroM0jXsILe",
    "outputId": "c240708c-b497-42a0-fa27-b757cd593532"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score, roc_curve, auc\n",
    "\n",
    "y_score = reloaded_rf.predict_proba(Test_X_Tfidf)\n",
    "rf_fpr, rf_tpr, thresholds = roc_curve(Test_Y, y_score[:,1], pos_label=1)\n",
    "area = auc(rf_fpr, rf_tpr)\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_theme()\n",
    "\n",
    "plt.figure(figsize=(8,6))\n",
    "lw = 2\n",
    "plt.plot(rf_fpr, rf_tpr, color='darkorange',\n",
    "         lw=lw, label='ROC curve (area = %0.2f)' % area)\n",
    "plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('ROC score RandomForest with Tf-Idf Features')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.savefig('ROC_RandomForest_tfidf.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 911,
     "status": "ok",
     "timestamp": 1622242298343,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "-WLRkYhJLBEX",
    "outputId": "4976873d-87b0-48f2-90dc-a668bf58cc15"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(8532, 3)"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_df = pd.DataFrame([rf_fpr, rf_tpr]).T\n",
    "roc_df.columns = ['rf_fpr', 'rf_tpr']\n",
    "roc_df['feat'] = 'tfidf'\n",
    "roc_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "executionInfo": {
     "elapsed": 186,
     "status": "ok",
     "timestamp": 1622242368757,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "7kw_Qo8JLBL4"
   },
   "outputs": [],
   "source": [
    "roc_df.to_csv('roc_tfidf_100k.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ChXEU2HSLiJ2"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Mju0JwQbkkGq"
   },
   "outputs": [],
   "source": [
    "reloaded_rf = pickle.load(open('drive/MyDrive/Milestone2/trained_models/rf_tfidf_10w.pkl', 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 4756,
     "status": "ok",
     "timestamp": 1622133456899,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "zh_RCTrqmaH1",
    "outputId": "6f11e0bc-fa1b-4c7d-f643-df6ac56def22"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64.98, 62.079047103410936)"
      ]
     },
     "execution_count": 16,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf_preds = reloaded_rf.predict(Test_X_Tfidf)\n",
    "rf_acc = accuracy_score(Test_Y, rf_preds)*100\n",
    "rf_f1 = f1_score(Test_Y, rf_preds)*100\n",
    "rf_acc, rf_f1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0lRATsxX9YXw"
   },
   "source": [
    "# Summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 111
    },
    "executionInfo": {
     "elapsed": 151,
     "status": "ok",
     "timestamp": 1622133478090,
     "user": {
      "displayName": "Ai Zhong",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Gh2MWN-q_WG3Xb3YELMkck5bLErFAYBIGqZ_8lQ=s64",
      "userId": "11851498678274176197"
     },
     "user_tz": 420
    },
    "id": "-15ZIIqMmata",
    "outputId": "6bf1cde9-87ce-4d75-c7b1-f409f6496dae"
   },
   "outputs": [
    {
     "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>accuracy</th>\n",
       "      <th>f1 score</th>\n",
       "      <th>model</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>63.77</td>\n",
       "      <td>63.777245</td>\n",
       "      <td>MLP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>64.98</td>\n",
       "      <td>62.079047</td>\n",
       "      <td>RandomForest</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   accuracy   f1 score         model\n",
       "0     63.77  63.777245           MLP\n",
       "1     64.98  62.079047  RandomForest"
      ]
     },
     "execution_count": 18,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "score_df = pd.DataFrame([[mlp_acc, rf_acc],[mlp_f1, rf_f1]]).T\n",
    "score_df.columns=['accuracy', 'f1 score']\n",
    "score_df['model'] = ['MLP','RandomForest']\n",
    "score_df.to_csv('scores_10w_tfidf.csv')\n",
    "score_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "DZA_GtMBsTUd"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "authorship_tag": "ABX9TyNvUkSv8g9UxIxWQjEvWkFs",
   "mount_file_id": "1HuxgHhAKLUbO25QbFDGSOETte_-QCxQe",
   "name": "Milestone2_TFIDF_100k.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}