{"metadata":{"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":50160,"databundleVersionId":7602123,"sourceType":"competition"}],"dockerImageVersionId":30646,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":true},"kernelspec":{"name":"python3","display_name":"Python 3","language":"python"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"papermill":{"default_parameters":{},"duration":9468.936222,"end_time":"2024-02-18T04:41:03.134316","environment_variables":{},"exception":null,"input_path":"__notebook__.ipynb","output_path":"__notebook__.ipynb","parameters":{},"start_time":"2024-02-18T02:03:14.198094","version":"2.5.0"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# 忽略未来警告信息\nimport warnings\nwarnings.simplefilter(action='ignore', category=FutureWarning)\n\n# 导入操作系统接口\nimport os\n# 导入垃圾回收模块\nimport gc\n\n# 导入NumPy库并重命名为np\nimport numpy as np\n# 导入Pandas库并重命名为pd\nimport pandas as pd\n\n# 导入Polars库并打印版本号\nimport polars as pl\nprint(pl.__version__)\n\n# 从glob模块导入glob函数，用于文件路径模式匹配\nfrom glob import glob\n# 从pathlib模块导入Path类，用于处理文件路径\nfrom pathlib import Path\n# 从datetime模块导入datetime类，用于处理日期和时间\nfrom datetime import datetime\n\n# 导入matplotlib的绘图模块并重命名为plt\nimport matplotlib.pyplot as plt\n# 导入seaborn库，用于数据可视化\nimport seaborn as sns\n\n# 从sklearn.model_selection导入时间序列分割类\nfrom sklearn.model_selection import TimeSeriesSplit, GroupKFold, StratifiedGroupKFold\n# 从sklearn.base导入基类Estimator和回归器混入类\nfrom sklearn.base import BaseEstimator, RegressorMixin\n# ROC-AUC计算方法\nfrom sklearn.metrics import roc_auc_score\n\n# 导入LightGBM库并重命名为lgb\nimport lightgbm as lgb","metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","papermill":{"duration":3.923946,"end_time":"2024-02-18T02:03:21.017083","exception":false,"start_time":"2024-02-18T02:03:17.093137","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Pipeline:\n    # 设置表格数据类型\n    # https://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/data\n    @staticmethod\n    def set_table_dtypes(df):\n        for col in df.columns:\n            if col in [\"case_id\", \"WEEK_NUM\", \"num_group1\", \"num_group2\"]:\n                df = df.with_columns(pl.col(col).cast(pl.Int64))\n            elif col in [\"date_decision\"]:\n                df = df.with_columns(pl.col(col).cast(pl.Date))\n            elif col[-1] in (\"P\", \"A\"):\n                df = df.with_columns(pl.col(col).cast(pl.Float64))\n            elif col[-1] in (\"M\",):\n                df = df.with_columns(pl.col(col).cast(pl.String))\n            elif col[-1] in (\"D\",):\n                df = df.with_columns(pl.col(col).cast(pl.Date))\n        return df\n\n    # 处理日期\n    @staticmethod\n    def handle_dates(df):\n        for col in df.columns:\n            if col[-1] in (\"D\",):\n                df = df.with_columns(pl.col(col) - pl.col(\"date_decision\"))\n                df = df.with_columns(pl.col(col).dt.total_days())\n        df = df.drop(\"date_decision\", \"MONTH\")\n\n        return df\n\n    # 过滤列\n    # 如果列名不在保留列表中，并且列的空值比例大于0.95，则删除该列\n    # 如果列名不在保留列表中，并且列的数据类型是String，且唯一值数量为1或大于200，则删除该列\n    @staticmethod\n    def filter_cols(df):\n        for col in df.columns:\n            if col not in [\"target\", \"case_id\", \"WEEK_NUM\"]:\n                isnull = df[col].is_null().mean()\n\n                if isnull > 0.95:\n                    df = df.drop(col)\n\n        for col in df.columns:\n            if (col not in [\"target\", \"case_id\", \"WEEK_NUM\"]) & (df[col].dtype == pl.String):\n                freq = df[col].n_unique()\n\n                if (freq == 1) | (freq > 200):\n                    df = df.drop(col)\n\n        return df","metadata":{"papermill":{"duration":0.02219,"end_time":"2024-02-18T02:03:21.046154","exception":false,"start_time":"2024-02-18T02:03:21.023964","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 定义了一个名为 Aggregator 的类，其中包含多个静态方法，用于生成聚合表达式\nclass Aggregator:\n    # 为数值类型的列生成最大值聚合表达式\n    @staticmethod\n    def num_expr(df):\n        cols = [col for col in df.columns if col[-1] in (\"P\", \"A\")]\n        expr_max = [pl.max(col).alias(f\"max_{col}\") for col in cols]\n        return expr_max\n\n    # 为日期类型的列生成最大值聚合表达式\n    @staticmethod\n    def date_expr(df):\n        cols = [col for col in df.columns if col[-1] in (\"D\",)]\n        expr_max = [pl.max(col).alias(f\"max_{col}\") for col in cols]\n        return expr_max\n\n    # 为字符串类型的列生成最大值聚合表达式\n    @staticmethod\n    def str_expr(df):\n        cols = [col for col in df.columns if col[-1] in (\"M\",)]\n        expr_max = [pl.max(col).alias(f\"max_{col}\") for col in cols]\n        return expr_max\n\n    # 为其他类型的列生成最大值聚合表达式\n    @staticmethod\n    def other_expr(df):\n        cols = [col for col in df.columns if col[-1] in (\"T\", \"L\")]\n        expr_max = [pl.max(col).alias(f\"max_{col}\") for col in cols]\n        return expr_max\n\n    # 为特定列（包含\"num_group\"）生成最大值聚合表达式\n    @staticmethod\n    def count_expr(df):\n        cols = [col for col in df.columns if \"num_group\" in col]\n        expr_max = [pl.max(col).alias(f\"max_{col}\") for col in cols]\n        return expr_max\n\n    # 获取所有类型的聚合表达式\n    @staticmethod\n    def get_exprs(df):\n        exprs = Aggregator.num_expr(df) + \\\n                Aggregator.date_expr(df) + \\\n                Aggregator.str_expr(df) + \\\n                Aggregator.other_expr(df) + \\\n                Aggregator.count_expr(df)\n        return exprs","metadata":{"papermill":{"duration":0.019165,"end_time":"2024-02-18T02:03:21.071306","exception":false,"start_time":"2024-02-18T02:03:21.052141","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 读取单个文件并进行预处理\ndef read_file(path, depth=None):\n    # 使用Polars库读取Parquet格式的文件\n    df = pl.read_parquet(path)\n    # 调用Pipeline类的set_table_dtypes方法设置数据类型\n    df = df.pipe(Pipeline.set_table_dtypes)\n    # 如果depth参数为1或2，对数据进行按\"case_id\"分组的聚合操作\n    if depth in [1, 2]:\n        df = df.group_by(\"case_id\").agg(Aggregator.get_exprs(df))\n    return df\n\n# 读取多个文件并进行预处理\ndef read_files(regex_path, depth=None):\n    chunks = []\n    for path in glob(str(regex_path)):\n        chunks.append(pl.read_parquet(path).pipe(Pipeline.set_table_dtypes))\n    df = pl.concat(chunks, how=\"vertical_relaxed\")\n    if depth in [1, 2]:\n        df = df.group_by(\"case_id\").agg(Aggregator.get_exprs(df))\n    return df","metadata":{"papermill":{"duration":0.016236,"end_time":"2024-02-18T02:03:21.093682","exception":false,"start_time":"2024-02-18T02:03:21.077446","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 特征工程函数，用于添加新特征和合并数据框\ndef feature_eng(df_base, depth_0, depth_1, depth_2):\n    df_base = (\n        df_base\n        .with_columns(\n            month_decision = pl.col(\"date_decision\").dt.month(),\n            weekday_decision = pl.col(\"date_decision\").dt.weekday(),\n        )\n    )\n    for i, df in enumerate(depth_0 + depth_1 + depth_2):\n        df_base = df_base.join(df, how=\"left\", on=\"case_id\", suffix=f\"_{i}\")\n    df_base = df_base.pipe(Pipeline.handle_dates)\n    return df_base","metadata":{"papermill":{"duration":0.015061,"end_time":"2024-02-18T02:03:21.114737","exception":false,"start_time":"2024-02-18T02:03:21.099676","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 将Polars数据框转换为Pandas数据框，并处理类别列\ndef to_pandas(df_data, cat_cols=None):\n    df_data = df_data.to_pandas()\n    if cat_cols is None:\n        cat_cols = list(df_data.select_dtypes(\"object\").columns)\n    df_data[cat_cols] = df_data[cat_cols].astype(\"category\")\n    return df_data, cat_cols","metadata":{"papermill":{"duration":0.013702,"end_time":"2024-02-18T02:03:21.134273","exception":false,"start_time":"2024-02-18T02:03:21.120571","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ROOT            = Path(\"/kaggle/input/home-credit-credit-risk-model-stability\")\n# ROOT            = Path(\"./input\")\n\nTRAIN_DIR       = ROOT / \"parquet_files\" / \"train\"\nTEST_DIR        = ROOT / \"parquet_files\" / \"test\"","metadata":{"papermill":{"duration":0.01447,"end_time":"2024-02-18T02:03:21.165934","exception":false,"start_time":"2024-02-18T02:03:21.151464","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 读取训练集的文件\ndata_store = {\n    \"df_base\": read_file(TRAIN_DIR / \"train_base.parquet\"),\n    \"depth_0\": [\n        read_file(TRAIN_DIR / \"train_static_cb_0.parquet\"),\n        read_files(TRAIN_DIR / \"train_static_0_*.parquet\"),\n    ],\n    \"depth_1\": [\n        read_files(TRAIN_DIR / \"train_applprev_1_*.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_tax_registry_a_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_tax_registry_b_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_tax_registry_c_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_credit_bureau_b_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_other_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_person_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_deposit_1.parquet\", 1),\n        read_file(TRAIN_DIR / \"train_debitcard_1.parquet\", 1),\n    ],\n    \"depth_2\": [\n        read_file(TRAIN_DIR / \"train_credit_bureau_b_2.parquet\", 2),\n    ]\n}","metadata":{"papermill":{"duration":22.271196,"end_time":"2024-02-18T02:03:43.454174","exception":false,"start_time":"2024-02-18T02:03:21.182978","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 对训练集进行特征工程\ndf_train = feature_eng(**data_store)\nprint(\"train data shape:\\t\", df_train.shape)","metadata":{"papermill":{"duration":6.172803,"end_time":"2024-02-18T02:03:49.633275","exception":false,"start_time":"2024-02-18T02:03:43.460472","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 读取测试集文件\ndata_store = {\n    \"df_base\": read_file(TEST_DIR / \"test_base.parquet\"),\n    \"depth_0\": [\n        read_file(TEST_DIR / \"test_static_cb_0.parquet\"),\n        read_files(TEST_DIR / \"test_static_0_*.parquet\"),\n    ],\n    \"depth_1\": [\n        read_files(TEST_DIR / \"test_applprev_1_*.parquet\", 1),\n        read_file(TEST_DIR / \"test_tax_registry_a_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_tax_registry_b_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_tax_registry_c_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_credit_bureau_b_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_other_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_person_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_deposit_1.parquet\", 1),\n        read_file(TEST_DIR / \"test_debitcard_1.parquet\", 1),\n    ],\n    \"depth_2\": [\n        read_file(TEST_DIR / \"test_credit_bureau_b_2.parquet\", 2),\n    ]\n}","metadata":{"papermill":{"duration":0.347863,"end_time":"2024-02-18T02:03:49.999943","exception":false,"start_time":"2024-02-18T02:03:49.652080","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 对测试集进行特征工程\ndf_test = feature_eng(**data_store)\nprint(\"test data shape:\\t\", df_test.shape)","metadata":{"papermill":{"duration":0.040418,"end_time":"2024-02-18T02:03:50.047284","exception":false,"start_time":"2024-02-18T02:03:50.006866","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 进行特征筛选\ndf_train = df_train.pipe(Pipeline.filter_cols)\ndf_test = df_test.select([col for col in df_train.columns if col != \"target\"])\n\nprint(\"train data shape:\\t\", df_train.shape)\nprint(\"test data shape:\\t\", df_test.shape)","metadata":{"papermill":{"duration":2.067345,"end_time":"2024-02-18T02:03:52.120929","exception":false,"start_time":"2024-02-18T02:03:50.053584","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 转换为pandas\ndf_train, cat_cols = to_pandas(df_train)\ndf_test, cat_cols = to_pandas(df_test, cat_cols)","metadata":{"papermill":{"duration":8.362466,"end_time":"2024-02-18T02:04:00.489650","exception":false,"start_time":"2024-02-18T02:03:52.127184","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"del data_store\n\ngc.collect()","metadata":{"papermill":{"duration":0.118213,"end_time":"2024-02-18T02:04:00.615687","exception":false,"start_time":"2024-02-18T02:04:00.497474","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# 自定义的投票模型，API与sklearn的类似\nclass VotingModel(BaseEstimator, RegressorMixin):\n    def __init__(self, estimators):\n        super().__init__()\n        self.estimators = estimators\n\n    def fit(self, X, y=None):\n        return self\n\n    def predict(self, X):\n        y_preds = [estimator.predict(X) for estimator in self.estimators]\n        return np.mean(y_preds, axis=0)\n\n    def predict_proba(self, X):\n        y_preds = [estimator.predict_proba(X) for estimator in self.estimators]\n        return np.mean(y_preds, axis=0)","metadata":{"papermill":{"duration":0.016782,"end_time":"2024-02-18T02:04:15.554576","exception":false,"start_time":"2024-02-18T02:04:15.537794","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X = df_train.drop(columns=[\"target\", \"case_id\",\"WEEK_NUM\"])\ny = df_train[\"target\"]\nweeks = df_train[\"WEEK_NUM\"]\n\n# 定义交叉验证\ncv = StratifiedGroupKFold(n_splits=5, shuffle=False)\n\n# lgb模型参数\nparams = {\n    \"boosting_type\": \"gbdt\",\n    \"objective\": \"binary\",\n    \"metric\": \"auc\",\n    \"max_depth\": 10,\n    \"learning_rate\": 0.05,\n    \"max_bin\": 255,\n    \"n_estimators\": 1200,\n    \"colsample_bytree\": 0.8,\n    \"colsample_bynode\": 0.8,\n    \"verbose\": -1,\n    \"random_state\": 42,\n    \"reg_alpha\": 0.1,\n    \"reg_lambda\": 10,\n    \"extra_trees\":True,\n    'num_leaves':64,\n    \"device\": \"gpu\",  # Uncomment if you want to use GPU for training\n}\n\nfitted_models = []\ncv_scores = []\n\n# 交叉验证训练模型\nfor idx_train, idx_valid in cv.split(X, y, groups=weeks):\n    X_train, y_train = X.iloc[idx_train], y.iloc[idx_train]\n    X_valid, y_valid = X.iloc[idx_valid], y.iloc[idx_valid]\n\n    print(\"Valid week range: \", (weeks.iloc[idx_valid].min(), weeks.iloc[idx_valid].max()))\n\n    model = lgb.LGBMClassifier(**params)\n    model.fit(\n        X_train, y_train,\n        eval_set=[(X_valid, y_valid)],\n        callbacks=[lgb.log_evaluation(50), lgb.early_stopping(50)]\n    )\n\n    fitted_models.append(model)\n\n    y_pred_valid = model.predict_proba(X_valid)[:, 1]\n    auc_score = roc_auc_score(y_valid, y_pred_valid)\n    cv_scores.append(auc_score)\n\nmodel = VotingModel(fitted_models)\nprint(\"CV AUC scores: \", cv_scores)\nprint(\"Average CV AUC score: \", sum(cv_scores) / len(cv_scores))","metadata":{"papermill":{"duration":9405.190593,"end_time":"2024-02-18T04:41:00.752298","exception":false,"start_time":"2024-02-18T02:04:15.561705","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_test = df_test.drop(columns=[\"WEEK_NUM\"])\nX_test = X_test.set_index(\"case_id\")\n\nlgb_pred = pd.Series(model.predict_proba(X_test)[:, 1], index=X_test.index)","metadata":{"execution":{"iopub.execute_input":"2024-02-18T04:41:00.788871Z","iopub.status.busy":"2024-02-18T04:41:00.788163Z","iopub.status.idle":"2024-02-18T04:41:01.005586Z","shell.execute_reply":"2024-02-18T04:41:01.003192Z"},"papermill":{"duration":0.241447,"end_time":"2024-02-18T04:41:01.008222","exception":false,"start_time":"2024-02-18T04:41:00.766775","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Submission","metadata":{"papermill":{"duration":0.014354,"end_time":"2024-02-18T04:41:01.036538","exception":false,"start_time":"2024-02-18T04:41:01.022184","status":"completed"},"tags":[]}},{"cell_type":"code","source":"df_subm = pd.read_csv(ROOT / \"sample_submission.csv\")\ndf_subm = df_subm.set_index(\"case_id\")\n\ndf_subm[\"score\"] = lgb_pred","metadata":{"execution":{"iopub.execute_input":"2024-02-18T04:41:01.066516Z","iopub.status.busy":"2024-02-18T04:41:01.066132Z","iopub.status.idle":"2024-02-18T04:41:01.099463Z","shell.execute_reply":"2024-02-18T04:41:01.098236Z"},"papermill":{"duration":0.051178,"end_time":"2024-02-18T04:41:01.101982","exception":false,"start_time":"2024-02-18T04:41:01.050804","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(\"Check null: \", df_subm[\"score\"].isnull().any())","metadata":{"execution":{"iopub.execute_input":"2024-02-18T04:41:01.133050Z","iopub.status.busy":"2024-02-18T04:41:01.132671Z","iopub.status.idle":"2024-02-18T04:41:01.141125Z","shell.execute_reply":"2024-02-18T04:41:01.140212Z"},"papermill":{"duration":0.026061,"end_time":"2024-02-18T04:41:01.143140","exception":false,"start_time":"2024-02-18T04:41:01.117079","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_subm.head()","metadata":{"execution":{"iopub.execute_input":"2024-02-18T04:41:01.174103Z","iopub.status.busy":"2024-02-18T04:41:01.173765Z","iopub.status.idle":"2024-02-18T04:41:01.204732Z","shell.execute_reply":"2024-02-18T04:41:01.203552Z"},"papermill":{"duration":0.049255,"end_time":"2024-02-18T04:41:01.206707","exception":false,"start_time":"2024-02-18T04:41:01.157452","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_subm.to_csv(\"submission.csv\")","metadata":{"execution":{"iopub.execute_input":"2024-02-18T04:41:01.238926Z","iopub.status.busy":"2024-02-18T04:41:01.238580Z","iopub.status.idle":"2024-02-18T04:41:01.252549Z","shell.execute_reply":"2024-02-18T04:41:01.251403Z"},"papermill":{"duration":0.033171,"end_time":"2024-02-18T04:41:01.254940","exception":false,"start_time":"2024-02-18T04:41:01.221769","status":"completed"},"tags":[]},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"- 库：Pandas/Polars、sklearn/lightgbm/xgboost、matplotlib/seaborn\n- 模型：树模型 + 集成学习\n- 硬件：Kaggle Notebook","metadata":{}},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}