Add files via upload

feature_extraction_1.ipynb

@@ -0,0 +1 @@
+{"cells":[{"cell_type":"code","execution_count":null,"metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","trusted":true},"outputs":[],"source":["import pandas as pd\n","import numpy as np\n","from catboost import CatBoostClassifier, Pool\n","from lightgbm import LGBMClassifier\n","from sklearn.model_selection import StratifiedKFold\n","from sklearn import metrics\n","from sklearn.preprocessing import LabelEncoder\n","import random\n","import warnings\n","from sklearn.model_selection import train_test_split \n","from sklearn.metrics import confusion_matrix\n","from sklearn.metrics import f1_score\n","from sklearn.metrics import precision_score\n","from sklearn.metrics import recall_score\n","import numpy as np # linear algebra\n","import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n","import numpy as np\n","import pandas as pd\n","import matplotlib.pyplot as plt\n","from sklearn.preprocessing import StandardScaler\n","\n","from sklearn.model_selection import train_test_split, StratifiedKFold\n","from imblearn.over_sampling import SMOTE\n","from imblearn.under_sampling import RandomUnderSampler\n","from sklearn.linear_model import LogisticRegressionCV, LogisticRegression\n","from sklearn.metrics import f1_score, precision_score, recall_score\n","import seaborn as sns\n","from sklearn.metrics import roc_curve, roc_auc_score, confusion_matrix, auc, f1_score\n","import tqdm\n","random.seed(42)\n","np.random.seed(42)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["df = pd.read_parquet('/kaggle/input/purple-hack/train_ai_comp_final_dp.parquet')\n","test = pd.read_parquet('/kaggle/input/purple-hack-with-test/test_sber.parquet')"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["for col in df.columns:\n","    if df[col].dtype == 'int64':\n","        df[col] = df[col].astype('int32')\n","    elif df[col].dtype == 'float64':\n","        df[col] = df[col].astype('float32')\n","    elif df[col].dtype == 'object':\n","        if len(df[col].unique()) / len(df[col]) < 0.5:\n","            df[col] = df[col].astype('category')"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["# Удаление фичей, которые коррелируют друг с другом больше, чем на 0.9\n","def get_correlated_feats(corr_matrix, feat_stats, greater_is_better=True, corr_threshold=0.95):\n","    cols = corr_matrix.columns.to_list()\n","    dropped = {col:0 for col in cols}\n","    for col in tqdm.tqdm(cols, desc='Get correlated features'):\n","        if dropped[col] == 0:\n","            columns_to_check = corr_matrix.index.values[np.abs(corr_matrix[col]) >= corr_threshold]\n","            if len(columns_to_check) > 1:\n","                if feat_stats is None:\n","                    bad_cols = columns_to_check[1:]\n","                else:\n","                    sel_stats = feat_stats.loc[columns_to_check]\n","                    if greater_is_better:\n","                        bad_cond = np.abs(sel_stats) < np.abs(sel_stats).max()\n","                    else:\n","                        bad_cond = np.abs(sel_stats) > np.abs(sel_stats).min()\n","                        \n","                    bad_cols = sel_stats[bad_cond].index.to_list()\n","                    norm_cols = sel_stats[~bad_cond].index.to_list()\n","                    if len(norm_cols) > 1:\n","                        for norm_col in norm_cols[1:]:\n","                            dropped[norm_col] += 1\n","                            \n","                for bad_col in bad_cols:\n","                    dropped[bad_col] += 1\n","    high_corr_cols = [c for c in dropped.keys() if dropped[c] > 0]\n","    return high_corr_cols"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["clean_df = df.copy()\n","nan_percentage = (clean_df == 0).mean()\n","cols_to_drop = list(nan_percentage[nan_percentage > 0.95].index)\n","\n","ignore_features = ['id', 'target', 'sample_ml_new', 'feature756']+cols_to_drop\n","print(len(ignore_features))\n","clean_df = clean_df.drop(columns=ignore_features)\n","\n","corr_mx = pd.DataFrame(np.corrcoef(clean_df.values, rowvar=False), columns=clean_df.columns, index=clean_df.columns)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["corr_feats = get_correlated_feats(corr_mx, feat_stats=nan_percentage, greater_is_better=False, corr_threshold=0.95)\n","len(corr_feats)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["df_no_corr = clean_df.drop(columns=corr_feats)\n","print(\"Было:\", df.shape)\n","print(\"Стало:\", df_no_corr.shape)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["final_feats = df_no_corr.columns\n","\n","X = df[final_feats].drop(columns = ['feature642'])\n","y = df['target']"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["X = X.reset_index(drop = True)\n","y = y.reset_index(drop = True)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["cat_cols = []\n","for col in X.columns:\n","    if len(X[col].value_counts()) < 3:\n","        cat_cols.append(col)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["X_train, X_test, y_train, y_test = train_test_split(X, y, stratify = y, test_size=0.25, random_state = 42)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["cat = CatBoostClassifier(random_seed = 42, class_weights = [1, 10])\n","cat.fit(X_train,y_train)"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["importance_df = pd.DataFrame({'Feature': X_train.columns, 'Importance': feature_importance})\n","\n","importance_df = importance_df.sort_values(by='Importance', ascending=False)\n","\n","top_10_features = importance_df.head(90)['Feature'].tolist()\n","\n","X_train_top_90 = X_train[top_90_features]\n","X_test_top_90 = X_test[top_90_features]"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":["X_train_top_90['sin341'] = np.sin(X_train_top_90['feature341'])\n","X_train_top_90['log940'] = np.log1p(X_train_top_90['feature940'])\n","X_train_top_90['new1'] = (X_train_top_90['feature1004'] * X_train_top_90['feature994'])\n","\n","X_test_top_90['sin341'] = np.sin(X_test_top_90['feature341'])\n","X_test_top_90['log940'] = np.log1p(X_test_top_90['feature940'])\n","X_test_top_90['new1'] = (X_test_top_90['feature1004'] * X_test_top_90['feature994'])"]}],"metadata":{"kaggle":{"accelerator":"none","dataSources":[],"isGpuEnabled":false,"isInternetEnabled":true,"language":"python","sourceType":"notebook"},"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":"3.6.4"}},"nbformat":4,"nbformat_minor":4}