train_XGBoost.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 训练XGBoost模型"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.metrics import classification_report,accuracy_score\n",
    "from sklearn.model_selection  import train_test_split\n",
    "import xgboost as xgb\n",
    "\n",
    "ZHFONT = matplotlib.font_manager.FontProperties(fname=\"data/SourceHanSansSC-Regular.otf\")\n",
    "\n",
    "df_train = pd.read_csv(\"data/train_new_2.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 划分数据集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 划分数据集\n",
    "targets = df_train.pop('target')\n",
    "\n",
    "train_data, val_data, train_targets, val_targets = train_test_split(df_train, targets, test_size = 0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 模型训练"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/root/miniconda3/envs/bi/lib/python3.13/site-packages/xgboost/core.py:723: FutureWarning: Pass `evals` as keyword args.\n",
      "  warnings.warn(msg, FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\teval-logloss:0.67779\n",
      "[1]\teval-logloss:0.66509\n",
      "[2]\teval-logloss:0.65456\n",
      "[3]\teval-logloss:0.64561\n",
      "[4]\teval-logloss:0.63801\n",
      "[5]\teval-logloss:0.63154\n",
      "[6]\teval-logloss:0.62601\n",
      "[7]\teval-logloss:0.62123\n",
      "[8]\teval-logloss:0.61705\n",
      "[9]\teval-logloss:0.61344\n",
      "[10]\teval-logloss:0.61013\n",
      "[11]\teval-logloss:0.60727\n",
      "[12]\teval-logloss:0.60482\n",
      "[13]\teval-logloss:0.60251\n",
      "[14]\teval-logloss:0.60051\n",
      "[15]\teval-logloss:0.59867\n",
      "[16]\teval-logloss:0.59703\n",
      "[17]\teval-logloss:0.59547\n",
      "[18]\teval-logloss:0.59403\n",
      "[19]\teval-logloss:0.59265\n",
      "[20]\teval-logloss:0.59145\n",
      "[21]\teval-logloss:0.59036\n",
      "[22]\teval-logloss:0.58913\n",
      "[23]\teval-logloss:0.58799\n",
      "[24]\teval-logloss:0.58696\n",
      "[25]\teval-logloss:0.58606\n",
      "[26]\teval-logloss:0.58515\n",
      "[27]\teval-logloss:0.58419\n",
      "[28]\teval-logloss:0.58341\n",
      "[29]\teval-logloss:0.58268\n",
      "[30]\teval-logloss:0.58178\n",
      "[31]\teval-logloss:0.58100\n",
      "[32]\teval-logloss:0.58044\n",
      "[33]\teval-logloss:0.57969\n",
      "[34]\teval-logloss:0.57902\n",
      "[35]\teval-logloss:0.57830\n",
      "[36]\teval-logloss:0.57773\n",
      "[37]\teval-logloss:0.57699\n",
      "[38]\teval-logloss:0.57638\n",
      "[39]\teval-logloss:0.57569\n",
      "[40]\teval-logloss:0.57523\n",
      "[41]\teval-logloss:0.57467\n",
      "[42]\teval-logloss:0.57429\n",
      "[43]\teval-logloss:0.57382\n",
      "[44]\teval-logloss:0.57329\n",
      "[45]\teval-logloss:0.57269\n",
      "[46]\teval-logloss:0.57230\n",
      "[47]\teval-logloss:0.57175\n",
      "[48]\teval-logloss:0.57122\n",
      "[49]\teval-logloss:0.57090\n",
      "[50]\teval-logloss:0.57052\n",
      "[51]\teval-logloss:0.56990\n",
      "[52]\teval-logloss:0.56945\n",
      "[53]\teval-logloss:0.56911\n",
      "[54]\teval-logloss:0.56866\n",
      "[55]\teval-logloss:0.56813\n",
      "[56]\teval-logloss:0.56747\n",
      "[57]\teval-logloss:0.56718\n",
      "[58]\teval-logloss:0.56679\n",
      "[59]\teval-logloss:0.56644\n",
      "[60]\teval-logloss:0.56621\n",
      "[61]\teval-logloss:0.56593\n",
      "[62]\teval-logloss:0.56548\n",
      "[63]\teval-logloss:0.56497\n",
      "[64]\teval-logloss:0.56448\n",
      "[65]\teval-logloss:0.56402\n",
      "[66]\teval-logloss:0.56376\n",
      "[67]\teval-logloss:0.56323\n",
      "[68]\teval-logloss:0.56276\n",
      "[69]\teval-logloss:0.56233\n",
      "[70]\teval-logloss:0.56215\n",
      "[71]\teval-logloss:0.56181\n",
      "[72]\teval-logloss:0.56137\n",
      "[73]\teval-logloss:0.56105\n",
      "[74]\teval-logloss:0.56071\n",
      "[75]\teval-logloss:0.56023\n",
      "[76]\teval-logloss:0.55990\n",
      "[77]\teval-logloss:0.55960\n",
      "[78]\teval-logloss:0.55924\n",
      "[79]\teval-logloss:0.55895\n",
      "[80]\teval-logloss:0.55879\n",
      "[81]\teval-logloss:0.55839\n",
      "[82]\teval-logloss:0.55824\n",
      "[83]\teval-logloss:0.55792\n",
      "[84]\teval-logloss:0.55775\n",
      "[85]\teval-logloss:0.55734\n",
      "[86]\teval-logloss:0.55696\n",
      "[87]\teval-logloss:0.55671\n",
      "[88]\teval-logloss:0.55637\n",
      "[89]\teval-logloss:0.55610\n",
      "[90]\teval-logloss:0.55585\n",
      "[91]\teval-logloss:0.55558\n",
      "[92]\teval-logloss:0.55512\n",
      "[93]\teval-logloss:0.55490\n",
      "[94]\teval-logloss:0.55455\n",
      "[95]\teval-logloss:0.55440\n",
      "[96]\teval-logloss:0.55421\n",
      "[97]\teval-logloss:0.55395\n",
      "[98]\teval-logloss:0.55365\n",
      "[99]\teval-logloss:0.55333\n"
     ]
    }
   ],
   "source": [
    "# 创建DMatrix数据结构\n",
    "dtrain = xgb.DMatrix(train_data, label=train_targets)\n",
    "dval = xgb.DMatrix(val_data, label=val_targets)\n",
    "\n",
    "# 设置模型超参数\n",
    "params = {\n",
    "    'objective': 'binary:logistic',\n",
    "    'eval_metric': 'logloss',\n",
    "    'learning_rate': 0.1,\n",
    "    'max_depth': 10,\n",
    "    'min_child_weight': 10,\n",
    "}\n",
    "EPOCH_NUM = 100\n",
    "\n",
    "# 训练模型\n",
    "evals_result = {}\n",
    "bst = xgb.train(params, dtrain, EPOCH_NUM, [(dval, 'eval')], evals_result=evals_result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'eval': OrderedDict({'logloss': [0.6777924848025322, 0.6650865077611997, 0.6545572597817488, 0.6456075131780832, 0.6380096073612208, 0.6315404513089156, 0.6260050314110844, 0.6212320238195251, 0.6170519322331659, 0.6134351673369792, 0.6101261181177707, 0.6072661320901327, 0.6048175199681247, 0.6025136207114723, 0.6005114196105716, 0.598673743426782, 0.5970273624195617, 0.595470492803495, 0.5940312153356146, 0.5926543955448115, 0.5914489778315136, 0.5903615699279994, 0.5891310223628571, 0.5879902097956273, 0.5869607969014095, 0.5860563933083958, 0.5851462614327885, 0.5841945642911935, 0.5834091897483034, 0.5826754248947262, 0.5817816706470313, 0.5810037701328289, 0.5804353838279157, 0.5796908810329863, 0.5790220333087831, 0.5783025872443359, 0.5777327739468381, 0.5769949818432525, 0.5763827260152616, 0.5756941277990425, 0.5752339481770802, 0.5746673565108421, 0.5742909825401066, 0.5738196619177964, 0.5732871228404379, 0.5726872075045396, 0.5722993179780199, 0.5717511010152005, 0.5712154803502404, 0.5708957350112269, 0.5705162674601291, 0.5698960090016667, 0.5694477926966255, 0.56911393015645, 0.5686621611602365, 0.5681342464403604, 0.5674727718304664, 0.5671786106644551, 0.5667899059376584, 0.5664394224114432, 0.5662052458115014, 0.5659253757238778, 0.5654849031678264, 0.5649703486715326, 0.5644812161604409, 0.564019564487387, 0.5637639460464474, 0.5632301521229225, 0.562757988015851, 0.5623280698767481, 0.5621545132679935, 0.5618051588365203, 0.561369199042571, 0.5610506182170095, 0.56071141494691, 0.5602286784207441, 0.5599031773086439, 0.559604698390565, 0.5592444705641229, 0.5589546472520346, 0.558785274081452, 0.5583937366638695, 0.5582401815813394, 0.5579197367276194, 0.5577499866822316, 0.55733978480944, 0.5569639080477651, 0.5567107132904561, 0.5563713937996707, 0.556104983038192, 0.5558519491108317, 0.5555752847047778, 0.5551179955546836, 0.5549041686928731, 0.5545509013062275, 0.5544035352355668, 0.5542092286267505, 0.5539504168474501, 0.5536461622645766, 0.5533252178168168]})}\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 可视化训练过程中的loss\n",
    "print(evals_result)\n",
    "# 获取logloss\n",
    "loss_result = evals_result['eval']['logloss']\n",
    "# 绘制损失折线图\n",
    "plt.figure(figsize=(12, 6))\n",
    "plt.plot(range(1, EPOCH_NUM + 1), loss_result)\n",
    "plt.title('模型损失变化',fontproperties=ZHFONT)\n",
    "plt.xlabel('训练轮次',fontproperties=ZHFONT)\n",
    "plt.ylabel('Logloss',fontproperties=ZHFONT)\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.71      0.73      0.72    366185\n",
      "           1       0.73      0.70      0.71    371400\n",
      "\n",
      "    accuracy                           0.72    737585\n",
      "   macro avg       0.72      0.72      0.72    737585\n",
      "weighted avg       0.72      0.72      0.72    737585\n",
      "\n",
      "AUC: 0.72\n"
     ]
    }
   ],
   "source": [
    "# 在val数据集上模型评估\n",
    "from sklearn.metrics import roc_auc_score\n",
    "pred = bst.predict(dval)\n",
    "predicted_classes = [1 if i > 0.5 else 0 for i in pred]\n",
    "\n",
    "# 计算准确率\n",
    "accuracy = classification_report(val_targets, predicted_classes)\n",
    "print(accuracy)\n",
    "# 计算AUC分数\n",
    "auc = roc_auc_score(val_targets, predicted_classes)\n",
    "print(f'AUC: {auc:.2f}')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "bi",
   "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": "3.13.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}