final.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import pandas as pd\n",
    "# import numpy as np\n",
    "# from sklearn.model_selection import train_test_split\n",
    "# from sklearn.preprocessing import StandardScaler\n",
    "# from sklearn.ensemble import GradientBoostingClassifier\n",
    "# from sklearn.metrics import accuracy_score, roc_auc_score, classification_report\n",
    "# import csv\n",
    "\n",
    "# # CLEANING FUNCTION FOR BOTH TRAIN AND TEST DATA\n",
    "# def clean_data(df):\n",
    "#     # 1. Handling missing values\n",
    "#     df = df.assign(\n",
    "#         job=df['job'].fillna('Unknown'),\n",
    "#         education=df['education'].fillna('Unknown'),\n",
    "#         contact=df['contact'].fillna('Missing'),\n",
    "#         poutcome=df['poutcome'].fillna('No previous contact')\n",
    "#     )\n",
    "\n",
    "#     # 2. Convert 'last contact date' to datetime\n",
    "#     df['last contact date'] = pd.to_datetime(df['last contact date'])\n",
    "\n",
    "#     # 3. Handle potential outliers in 'balance'\n",
    "#     Q1 = df['balance'].quantile(0.25)\n",
    "#     Q3 = df['balance'].quantile(0.75)\n",
    "#     IQR = Q3 - Q1\n",
    "#     lower_bound = Q1 - 1.5 * IQR\n",
    "#     upper_bound = Q3 + 1.5 * IQR\n",
    "#     df['balance'] = df['balance'].clip(lower_bound, upper_bound)\n",
    "\n",
    "#     # 4. Encode categorical variables\n",
    "#     categorical_cols = ['job', 'marital', 'education', 'default', 'housing', 'loan', 'contact', 'poutcome']\n",
    "#     df_encoded = pd.get_dummies(df, columns=categorical_cols, drop_first=True)\n",
    "\n",
    "#     # 5. Normalize numerical variables\n",
    "#     scaler = StandardScaler()\n",
    "#     numerical_cols = ['age', 'balance', 'duration', 'campaign', 'pdays', 'previous']\n",
    "#     df_encoded[numerical_cols] = scaler.fit_transform(df_encoded[numerical_cols])\n",
    "\n",
    "#     # 6. Handle 'pdays' variable\n",
    "#     df_encoded['pdays'] = df_encoded['pdays'].replace(-1, 1000)\n",
    "\n",
    "#     # 7. Create a new feature for the month of last contact\n",
    "#     df_encoded['last_contact_month'] = df['last contact date'].dt.month\n",
    "\n",
    "#     # 8. Drop the original 'last contact date' column\n",
    "#     df_encoded = df_encoded.drop('last contact date', axis=1)\n",
    "    \n",
    "#     return df_encoded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy: 0.8663\n",
      "ROC AUC: 0.9036\n",
      "\n",
      "Classification Report:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.90      0.95      0.92      6645\n",
      "           1       0.59      0.42      0.49      1198\n",
      "\n",
      "    accuracy                           0.87      7843\n",
      "   macro avg       0.74      0.69      0.71      7843\n",
      "weighted avg       0.85      0.87      0.86      7843\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# # Load the training data\n",
    "# df_train = pd.read_csv('train.csv')\n",
    "\n",
    "# # Clean the training data\n",
    "# df_train_clean = clean_data(df_train)\n",
    "\n",
    "# # Ensure the target variable is properly encoded\n",
    "# df_train_clean['target'] = df_train_clean['target'].map({'no': 0, 'yes': 1})\n",
    "\n",
    "# # Split the training data into training and validation sets\n",
    "# X = df_train_clean.drop('target', axis=1)\n",
    "# y = df_train_clean['target']\n",
    "# X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)\n",
    "\n",
    "# # Scale the features\n",
    "# scaler = StandardScaler()\n",
    "# X_train_scaled = scaler.fit_transform(X_train)\n",
    "# X_val_scaled = scaler.transform(X_val)\n",
    "\n",
    "# # Train the model\n",
    "# gb_model = GradientBoostingClassifier(n_estimators=100, random_state=42)\n",
    "# gb_model.fit(X_train_scaled, y_train)\n",
    "\n",
    "# # Evaluate the model\n",
    "# y_val_pred = gb_model.predict(X_val_scaled)\n",
    "# y_val_pred_proba = gb_model.predict_proba(X_val_scaled)[:, 1]\n",
    "# print(f\"Accuracy: {accuracy_score(y_val, y_val_pred):.4f}\")\n",
    "# print(f\"ROC AUC: {roc_auc_score(y_val, y_val_pred_proba):.4f}\")\n",
    "# print(\"\\nClassification Report:\")\n",
    "# print(classification_report(y_val, y_val_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission file created successfully!\n"
     ]
    }
   ],
   "source": [
    "# # TEST SET PREDICTION\n",
    "# df_test = pd.read_csv('test.csv')\n",
    "# df_test_clean = clean_data(df_test)\n",
    "\n",
    "# # Scale the test data (use the same scaler from training)\n",
    "# X_test_scaled = scaler.transform(df_test_clean)\n",
    "\n",
    "# # Generate predictions\n",
    "# test_preds = gb_model.predict(X_test_scaled)\n",
    "\n",
    "# # Convert predictions from numerical to 'yes'/'no' labels\n",
    "# test_preds_labels = np.where(test_preds == 1, 'yes', 'no')\n",
    "\n",
    "# # Prepare the submission dataframe\n",
    "# submission = pd.DataFrame({\n",
    "#     'id': df_test.index,  # Assuming test.csv contains an 'id' column\n",
    "#     'target': test_preds_labels\n",
    "# })\n",
    "\n",
    "# # Save to submissions.csv\n",
    "# submission.to_csv('submissions.csv', index=False, quoting=csv.QUOTE_NONNUMERIC)\n",
    "\n",
    "# print(\"Submission file created successfully!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<bound method NDFrame.head of        age  balance  duration  campaign  pdays  previous target  job_admin.  \\\n",
      "11002   35   4895.5       139         8    273         7     no       False   \n",
      "16008   54    384.0       200         1   1000         0    yes       False   \n",
      "24722   21    216.0       414         1   1000         0    yes       False   \n",
      "1722    32   4895.5       141         1   1000         0     no       False   \n",
      "21374   41    -62.0        55         2   1000         0     no        True   \n",
      "...    ...      ...       ...       ...    ...       ...    ...         ...   \n",
      "8964    52    428.0       153         1   1000         0     no       False   \n",
      "23322   28   4895.5      1143        28    708        84     no       False   \n",
      "3638    37   1580.0        92        12   1000         0     no       False   \n",
      "20677   28    423.0       245         1   1000         0     no        True   \n",
      "38130   28      0.0       269         1     40         2    yes       False   \n",
      "\n",
      "       job_blue-collar  job_entrepreneur  ...  education_tertiary  \\\n",
      "11002            False             False  ...               False   \n",
      "16008            False             False  ...                True   \n",
      "24722            False             False  ...               False   \n",
      "1722             False             False  ...                True   \n",
      "21374            False             False  ...               False   \n",
      "...                ...               ...  ...                 ...   \n",
      "8964             False             False  ...               False   \n",
      "23322            False              True  ...               False   \n",
      "3638             False             False  ...               False   \n",
      "20677            False             False  ...               False   \n",
      "38130            False             False  ...                True   \n",
      "\n",
      "       default_yes  housing_yes  loan_yes  contact_cellular  \\\n",
      "11002        False         True     False             False   \n",
      "16008        False        False      True              True   \n",
      "24722        False        False     False              True   \n",
      "1722         False        False     False             False   \n",
      "21374        False         True      True             False   \n",
      "...            ...          ...       ...               ...   \n",
      "8964         False         True     False             False   \n",
      "23322        False         True      True              True   \n",
      "3638         False        False     False              True   \n",
      "20677        False         True     False              True   \n",
      "38130        False         True     False              True   \n",
      "\n",
      "       contact_telephone  poutcome_failure  poutcome_other  poutcome_success  \\\n",
      "11002               True              True           False             False   \n",
      "16008              False             False           False             False   \n",
      "24722              False             False           False             False   \n",
      "1722               False             False           False             False   \n",
      "21374              False             False           False             False   \n",
      "...                  ...               ...             ...               ...   \n",
      "8964               False             False           False             False   \n",
      "23322              False             False            True             False   \n",
      "3638               False             False           False             False   \n",
      "20677              False             False           False             False   \n",
      "38130              False              True           False             False   \n",
      "\n",
      "       last_contact_month  \n",
      "11002                   2  \n",
      "16008                   5  \n",
      "24722                   8  \n",
      "1722                    5  \n",
      "21374                   5  \n",
      "...                   ...  \n",
      "8964                    5  \n",
      "23322                   1  \n",
      "3638                    8  \n",
      "20677                   7  \n",
      "38130                   5  \n",
      "\n",
      "[31368 rows x 32 columns]>\n",
      "Accuracy: 0.8730\n",
      "ROC AUC: 0.9079\n",
      "\n",
      "Classification Report:\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.91      0.95      0.93      6677\n",
      "           1       0.60      0.45      0.51      1166\n",
      "\n",
      "    accuracy                           0.87      7843\n",
      "   macro avg       0.75      0.70      0.72      7843\n",
      "weighted avg       0.86      0.87      0.87      7843\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# import pandas as pd\n",
    "# import numpy as np\n",
    "# from sklearn.model_selection import train_test_split\n",
    "# from sklearn.preprocessing import StandardScaler\n",
    "# from sklearn.ensemble import GradientBoostingClassifier\n",
    "# from sklearn.metrics import accuracy_score, roc_auc_score, classification_report\n",
    "# import csv\n",
    "\n",
    "# # CLEANING FUNCTION FOR BOTH TRAIN AND TEST DATA\n",
    "# def clean_data(df):\n",
    "#     # 1. Handling missing values\n",
    "#     df = df.assign(\n",
    "#         job=df['job'].fillna('Unknown'),\n",
    "#         education=df['education'].fillna('Unknown'),\n",
    "#         contact=df['contact'].fillna('Missing'),\n",
    "#         poutcome=df['poutcome'].fillna('No previous contact')\n",
    "#     )\n",
    "\n",
    "#     # 2. Convert 'last contact date' to datetime\n",
    "#     df['last contact date'] = pd.to_datetime(df['last contact date'])\n",
    "\n",
    "#     # 3. Handle potential outliers in 'balance'\n",
    "#     Q1 = df['balance'].quantile(0.25)\n",
    "#     Q3 = df['balance'].quantile(0.75)\n",
    "#     IQR = Q3 - Q1\n",
    "#     lower_bound = Q1 - 1.5 * IQR\n",
    "#     upper_bound = Q3 + 1.5 * IQR\n",
    "#     df['balance'] = df['balance'].clip(lower_bound, upper_bound)\n",
    "\n",
    "#     # 4. Encode categorical variables\n",
    "#     categorical_cols = ['job', 'marital', 'education', 'default', 'housing', 'loan', 'contact', 'poutcome']\n",
    "#     df_encoded = pd.get_dummies(df, columns=categorical_cols, drop_first=True)\n",
    "\n",
    "#     # 5. Handle 'pdays' variable\n",
    "#     df_encoded['pdays'] = df_encoded['pdays'].replace(-1, 1000)\n",
    "\n",
    "#     # 6. Create a new feature for the month of last contact\n",
    "#     df_encoded['last_contact_month'] = df['last contact date'].dt.month\n",
    "\n",
    "#     # 7. Drop the original 'last contact date' column\n",
    "#     df_encoded = df_encoded.drop('last contact date', axis=1)\n",
    "    \n",
    "#     return df_encoded\n",
    "\n",
    "# # Load the dataset\n",
    "# df = pd.read_csv('train.csv')\n",
    "\n",
    "# # Split the dataset first (before cleaning)\n",
    "# df_train, df_val = train_test_split(df, test_size=0.2, random_state=42, stratify=df['target'])\n",
    "\n",
    "# # Clean the training data\n",
    "# df_train_clean = clean_data(df_train)\n",
    "\n",
    "# # Clean the validation data\n",
    "# df_val_clean = clean_data(df_val)\n",
    "\n",
    "# # Ensure the target variable is properly encoded\n",
    "# df_train_clean['target'] = df_train_clean['target'].map({'no': 0, 'yes': 1})\n",
    "# df_val_clean['target'] = df_val_clean['target'].map({'no': 0, 'yes': 1})\n",
    "\n",
    "# # Separate features and target variable for both training and validation sets\n",
    "# X_train = df_train_clean.drop('target', axis=1)\n",
    "# y_train = df_train_clean['target']\n",
    "# X_val = df_val_clean.drop('target', axis=1)\n",
    "# y_val = df_val_clean['target']\n",
    "\n",
    "# # Scale the features\n",
    "# scaler = StandardScaler()\n",
    "# X_train_scaled = scaler.fit_transform(X_train)\n",
    "# X_val_scaled = scaler.transform(X_val)\n",
    "\n",
    "# # Train the model\n",
    "# gb_model = GradientBoostingClassifier(n_estimators=100, random_state=42)\n",
    "# gb_model.fit(X_train_scaled, y_train)\n",
    "\n",
    "# # Evaluate the model on the validation set\n",
    "# y_val_pred = gb_model.predict(X_val_scaled)\n",
    "# y_val_pred_proba = gb_model.predict_proba(X_val_scaled)[:, 1]\n",
    "# print(f\"Accuracy: {accuracy_score(y_val, y_val_pred):.4f}\")\n",
    "# print(f\"ROC AUC: {roc_auc_score(y_val, y_val_pred_proba):.4f}\")\n",
    "# print(\"\\nClassification Report:\")\n",
    "# print(classification_report(y_val, y_val_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Submission file created successfully!\n"
     ]
    }
   ],
   "source": [
    "# # TEST SET PREDICTION\n",
    "# df_test = pd.read_csv('test.csv')\n",
    "# df_test_clean = clean_data(df_test)\n",
    "\n",
    "# # Scale the test data (use the same scaler from training)\n",
    "# X_test_scaled = scaler.transform(df_test_clean)\n",
    "\n",
    "# # Generate predictions\n",
    "# test_preds = gb_model.predict(X_test_scaled)\n",
    "\n",
    "# # Convert predictions from numerical to 'yes'/'no' labels\n",
    "# test_preds_labels = np.where(test_preds == 1, 'yes', 'no')\n",
    "\n",
    "# # Prepare the submission dataframe\n",
    "# submission = pd.DataFrame({\n",
    "#     'id': df_test.index,  # Assuming test.csv contains an 'id' column\n",
    "#     'target': test_preds_labels\n",
    "# })\n",
    "\n",
    "# # Save to submissions.csv\n",
    "# submission.to_csv('submission-yet-again.csv', index=False, quoting=csv.QUOTE_NONNUMERIC)\n",
    "\n",
    "# print(\"Submission file created successfully!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Columns in X_train: {'education_primary', 'contact_cellular', 'poutcome_No previous contact', 'default_no', 'job_unemployed', 'housing_no', 'last_contact_month_5', 'marital_married', 'previous', 'education_Unknown', 'job_blue-collar', 'default_yes', 'last_contact_month_1', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'duration', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'marital_divorced', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_Unknown', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'campaign', 'last_contact_month_2', 'age', 'pdays', 'balance', 'loan_no', 'poutcome_other', 'housing_yes', 'contact_Missing', 'last_contact_month_3', 'job_housemaid'}\n",
      "Columns in X_val: {'education_primary', 'contact_cellular', 'poutcome_No previous contact', 'default_no', 'job_unemployed', 'housing_no', 'last_contact_month_5', 'marital_married', 'previous', 'education_Unknown', 'job_blue-collar', 'default_yes', 'last_contact_month_1', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'duration', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'marital_divorced', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_Unknown', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'campaign', 'last_contact_month_2', 'age', 'pdays', 'balance', 'loan_no', 'poutcome_other', 'housing_yes', 'contact_Missing', 'last_contact_month_3', 'job_housemaid'}\n",
      "Numeric features: ['age', 'balance', 'duration', 'campaign', 'pdays', 'previous']\n",
      "Categorical features: ['education_primary', 'contact_cellular', 'poutcome_No previous contact', 'default_no', 'job_unemployed', 'housing_no', 'last_contact_month_5', 'marital_married', 'education_Unknown', 'job_blue-collar', 'default_yes', 'last_contact_month_1', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'marital_divorced', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_Unknown', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'last_contact_month_2', 'loan_no', 'poutcome_other', 'housing_yes', 'contact_Missing', 'last_contact_month_3', 'job_housemaid']\n",
      "\n",
      "Data types:\n",
      "age                             int64\n",
      "balance                         int64\n",
      "duration                        int64\n",
      "campaign                        int64\n",
      "pdays                           int64\n",
      "previous                        int64\n",
      "job_Unknown                      bool\n",
      "job_admin.                       bool\n",
      "job_blue-collar                  bool\n",
      "job_entrepreneur                 bool\n",
      "job_housemaid                    bool\n",
      "job_management                   bool\n",
      "job_retired                      bool\n",
      "job_self-employed                bool\n",
      "job_services                     bool\n",
      "job_student                      bool\n",
      "job_technician                   bool\n",
      "job_unemployed                   bool\n",
      "marital_divorced                 bool\n",
      "marital_married                  bool\n",
      "marital_single                   bool\n",
      "education_Unknown                bool\n",
      "education_primary                bool\n",
      "education_secondary              bool\n",
      "education_tertiary               bool\n",
      "default_no                       bool\n",
      "default_yes                      bool\n",
      "housing_no                       bool\n",
      "housing_yes                      bool\n",
      "loan_no                          bool\n",
      "loan_yes                         bool\n",
      "contact_Missing                  bool\n",
      "contact_cellular                 bool\n",
      "contact_telephone                bool\n",
      "poutcome_No previous contact     bool\n",
      "poutcome_failure                 bool\n",
      "poutcome_other                   bool\n",
      "poutcome_success                 bool\n",
      "last_contact_month_1             bool\n",
      "last_contact_month_2             bool\n",
      "last_contact_month_3             bool\n",
      "last_contact_month_4             bool\n",
      "last_contact_month_5             bool\n",
      "last_contact_month_6             bool\n",
      "last_contact_month_7             bool\n",
      "last_contact_month_8             bool\n",
      "last_contact_month_9             bool\n",
      "last_contact_month_10            bool\n",
      "last_contact_month_11            bool\n",
      "last_contact_month_12            bool\n",
      "dtype: object\n",
      "\n",
      "Null values:\n",
      "age                             0\n",
      "balance                         0\n",
      "duration                        0\n",
      "campaign                        0\n",
      "pdays                           0\n",
      "previous                        0\n",
      "job_Unknown                     0\n",
      "job_admin.                      0\n",
      "job_blue-collar                 0\n",
      "job_entrepreneur                0\n",
      "job_housemaid                   0\n",
      "job_management                  0\n",
      "job_retired                     0\n",
      "job_self-employed               0\n",
      "job_services                    0\n",
      "job_student                     0\n",
      "job_technician                  0\n",
      "job_unemployed                  0\n",
      "marital_divorced                0\n",
      "marital_married                 0\n",
      "marital_single                  0\n",
      "education_Unknown               0\n",
      "education_primary               0\n",
      "education_secondary             0\n",
      "education_tertiary              0\n",
      "default_no                      0\n",
      "default_yes                     0\n",
      "housing_no                      0\n",
      "housing_yes                     0\n",
      "loan_no                         0\n",
      "loan_yes                        0\n",
      "contact_Missing                 0\n",
      "contact_cellular                0\n",
      "contact_telephone               0\n",
      "poutcome_No previous contact    0\n",
      "poutcome_failure                0\n",
      "poutcome_other                  0\n",
      "poutcome_success                0\n",
      "last_contact_month_1            0\n",
      "last_contact_month_2            0\n",
      "last_contact_month_3            0\n",
      "last_contact_month_4            0\n",
      "last_contact_month_5            0\n",
      "last_contact_month_6            0\n",
      "last_contact_month_7            0\n",
      "last_contact_month_8            0\n",
      "last_contact_month_9            0\n",
      "last_contact_month_10           0\n",
      "last_contact_month_11           0\n",
      "last_contact_month_12           0\n",
      "dtype: int64\n",
      "\n",
      "Approach 1: Modified ColumnTransformer\n",
      "Pipeline fitted successfully\n",
      "Validation Accuracy: 0.8725\n",
      "Validation ROC AUC: 0.9093\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split, cross_val_score\n",
    "from sklearn.preprocessing import StandardScaler, OneHotEncoder\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.metrics import accuracy_score, roc_auc_score, classification_report\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.compose import ColumnTransformer\n",
    "import csv\n",
    "\n",
    "# CLEANING FUNCTION FOR BOTH TRAIN AND TEST DATA\n",
    "def clean_data(df, is_train=True, train_columns=None):\n",
    "    # 1. Handling missing values\n",
    "    df = df.assign(\n",
    "        job=df['job'].fillna('Unknown'),\n",
    "        education=df['education'].fillna('Unknown'),\n",
    "        contact=df['contact'].fillna('Missing'),\n",
    "        poutcome=df['poutcome'].fillna('No previous contact')\n",
    "    )\n",
    "    \n",
    "    # 2. Convert 'last contact date' to datetime\n",
    "    df['last contact date'] = pd.to_datetime(df['last contact date'])\n",
    "    \n",
    "    # 3. Handle 'pdays' variable\n",
    "    df['pdays'] = df['pdays'].replace(-1, 1000)\n",
    "    \n",
    "    # 4. Create month categories\n",
    "    df['last_contact_month'] = df['last contact date'].dt.month\n",
    "    \n",
    "    # 5. Drop the original 'last contact date' column\n",
    "    df = df.drop('last contact date', axis=1)\n",
    "    \n",
    "    # 6. Encode categorical variables\n",
    "    categorical_cols = ['job', 'marital', 'education', 'default', 'housing', 'loan', 'contact', 'poutcome', 'last_contact_month']\n",
    "    df_encoded = pd.get_dummies(df, columns=categorical_cols, drop_first=False)\n",
    "    \n",
    "    if train_columns is not None:\n",
    "        # Add missing columns\n",
    "        for col in train_columns:\n",
    "            if col not in df_encoded.columns:\n",
    "                df_encoded[col] = 0\n",
    "        # Remove extra columns\n",
    "        df_encoded = df_encoded[train_columns]\n",
    "    \n",
    "    # 7. Only perform operations that might leak information during training\n",
    "    if is_train:\n",
    "        # Handle potential outliers in 'balance'\n",
    "        Q1 = df['balance'].quantile(0.25)\n",
    "        Q3 = df['balance'].quantile(0.75)\n",
    "        IQR = Q3 - Q1\n",
    "        lower_bound = Q1 - 1.5 * IQR\n",
    "        upper_bound = Q3 + 1.5 * IQR\n",
    "        df_encoded['balance'] = df_encoded['balance'].clip(lower_bound, upper_bound)\n",
    "    \n",
    "    return df_encoded\n",
    "\n",
    "# Load the dataset\n",
    "df = pd.read_csv('train.csv')\n",
    "\n",
    "# Clean the data before splitting\n",
    "df_clean = clean_data(df)\n",
    "\n",
    "# Split the cleaned dataset\n",
    "df_train, df_val = train_test_split(df_clean, test_size=0.2, random_state=42, stratify=df_clean['target'])\n",
    "\n",
    "# Ensure the target variable is properly encoded\n",
    "df_train['target'] = df_train['target'].map({'no': 0, 'yes': 1})\n",
    "df_val['target'] = df_val['target'].map({'no': 0, 'yes': 1})\n",
    "\n",
    "# Separate features and target variable for both training and validation sets\n",
    "X_train = df_train.drop('target', axis=1)\n",
    "y_train = df_train['target']\n",
    "X_val = df_val.drop('target', axis=1)\n",
    "y_val = df_val['target']\n",
    "\n",
    "# Print column information\n",
    "print(\"Columns in X_train:\", set(X_train.columns))\n",
    "print(\"Columns in X_val:\", set(X_val.columns))\n",
    "\n",
    "# Identify numeric and categorical columns\n",
    "all_columns = set(X_train.columns)\n",
    "numeric_features = ['age', 'balance', 'duration', 'campaign', 'pdays', 'previous']\n",
    "numeric_features = [col for col in numeric_features if col in all_columns]\n",
    "categorical_features = [col for col in all_columns if col not in numeric_features]\n",
    "\n",
    "print(\"Numeric features:\", numeric_features)\n",
    "print(\"Categorical features:\", categorical_features)\n",
    "\n",
    "# Check data types and null values\n",
    "print(\"\\nData types:\")\n",
    "print(X_train.dtypes)\n",
    "print(\"\\nNull values:\")\n",
    "print(X_train.isnull().sum())\n",
    "\n",
    "# Approach 1: Modified ColumnTransformer\n",
    "print(\"\\nApproach 1: Modified ColumnTransformer\")\n",
    "preprocessor = ColumnTransformer(\n",
    "    transformers=[\n",
    "        ('num', StandardScaler(), numeric_features),\n",
    "        ('cat', 'passthrough', categorical_features)\n",
    "    ])\n",
    "\n",
    "pipeline = Pipeline([\n",
    "    ('preprocessor', preprocessor),\n",
    "    ('classifier', GradientBoostingClassifier(n_estimators=100, max_depth=3, min_samples_leaf=5, learning_rate=0.1, random_state=42))\n",
    "])\n",
    "\n",
    "try:\n",
    "    pipeline.fit(X_train, y_train)\n",
    "    print(\"Pipeline fitted successfully\")\n",
    "    \n",
    "    # Make predictions on validation set\n",
    "    y_val_pred = pipeline.predict(X_val)\n",
    "    y_val_pred_proba = pipeline.predict_proba(X_val)[:, 1]\n",
    "    \n",
    "    # Evaluate the model on the validation set\n",
    "    print(f\"Validation Accuracy: {accuracy_score(y_val, y_val_pred):.4f}\")\n",
    "    print(f\"Validation ROC AUC: {roc_auc_score(y_val, y_val_pred_proba):.4f}\")\n",
    "except ValueError as e:\n",
    "    print(\"Error:\", str(e))\n",
    "\n",
    "# # Approach 2: Fitting without pipeline\n",
    "# print(\"\\nApproach 2: Fitting without pipeline\")\n",
    "# X_train_scaled = StandardScaler().fit_transform(X_train[numeric_features])\n",
    "# X_train_processed = np.hstack([X_train_scaled, X_train[categorical_features]])\n",
    "\n",
    "# model = GradientBoostingClassifier(n_estimators=100, max_depth=3, min_samples_leaf=5, learning_rate=0.1, random_state=42)\n",
    "\n",
    "# try:\n",
    "#     model.fit(X_train_processed, y_train)\n",
    "#     print(\"Model fitted successfully\")\n",
    "    \n",
    "#     # Prepare validation data\n",
    "#     X_val_scaled = StandardScaler().fit_transform(X_val[numeric_features])\n",
    "#     X_val_processed = np.hstack([X_val_scaled, X_val[categorical_features]])\n",
    "    \n",
    "#     # Make predictions\n",
    "#     y_val_pred = model.predict(X_val_processed)\n",
    "#     y_val_pred_proba = model.predict_proba(X_val_processed)[:, 1]\n",
    "    \n",
    "#     # Evaluate the model\n",
    "#     print(f\"Validation Accuracy: {accuracy_score(y_val, y_val_pred):.4f}\")\n",
    "#     print(f\"Validation ROC AUC: {roc_auc_score(y_val, y_val_pred_proba):.4f}\")\n",
    "# except ValueError as e:\n",
    "#     print(\"Error:\", str(e))\n",
    "\n",
    "# # Approach 3: Numeric features only\n",
    "# print(\"\\nApproach 3: Numeric features only\")\n",
    "# X_train_numeric = X_train[numeric_features]\n",
    "# model_numeric = GradientBoostingClassifier(n_estimators=100, max_depth=3, min_samples_leaf=5, learning_rate=0.1, random_state=42)\n",
    "\n",
    "# try:\n",
    "#     model_numeric.fit(X_train_numeric, y_train)\n",
    "#     print(\"Model fitted successfully with numeric features only\")\n",
    "    \n",
    "#     # Make predictions\n",
    "#     X_val_numeric = X_val[numeric_features]\n",
    "#     y_val_pred = model_numeric.predict(X_val_numeric)\n",
    "#     y_val_pred_proba = model_numeric.predict_proba(X_val_numeric)[:, 1]\n",
    "    \n",
    "#     # Evaluate the model\n",
    "#     print(f\"Validation Accuracy: {accuracy_score(y_val, y_val_pred):.4f}\")\n",
    "#     print(f\"Validation ROC AUC: {roc_auc_score(y_val, y_val_pred_proba):.4f}\")\n",
    "# except ValueError as e:\n",
    "#     print(\"Error:\", str(e))\n",
    "\n",
    "# # Approach 4: Adding categorical features one by one\n",
    "# print(\"\\nApproach 4: Adding categorical features one by one\")\n",
    "# for cat_feature in categorical_features:\n",
    "#     print(f\"Testing with feature: {cat_feature}\")\n",
    "#     X_train_test = pd.concat([X_train_numeric, X_train[cat_feature]], axis=1)\n",
    "#     model_test = GradientBoostingClassifier(n_estimators=100, max_depth=3, min_samples_leaf=5, learning_rate=0.1, random_state=42)\n",
    "#     try:\n",
    "#         model_test.fit(X_train_test, y_train)\n",
    "#         print(\"  Success\")\n",
    "#     except ValueError as e:\n",
    "#         print(f\"  Error: {str(e)}\")\n",
    "\n",
    "# # If all approaches work, proceed with test set prediction\n",
    "# try:\n",
    "#     # Load and clean test data\n",
    "#     df_test = pd.read_csv('test.csv')\n",
    "#     df_test_clean = clean_data(df_test, is_train=False, train_columns=X_train.columns)\n",
    "\n",
    "#     # Generate predictions using the pipeline (if it worked)\n",
    "#     test_preds_proba = pipeline.predict_proba(df_test_clean)[:, 1]\n",
    "\n",
    "#     # Convert probabilities to labels (adjust threshold if needed)\n",
    "#     test_preds_labels = np.where(test_preds_proba > 0.5, 'yes', 'no')\n",
    "\n",
    "#     # Prepare the submission dataframe\n",
    "#     submission = pd.DataFrame({\n",
    "#         'id': df_test.index,\n",
    "#         'target': test_preds_labels\n",
    "#     })\n",
    "\n",
    "    # Save to submissions.csv\n",
    "    submission.to_csv('submission.csv', index=False, quoting=csv.QUOTE_NONNUMERIC)\n",
    "\n",
    "    print(\"Submission file created successfully!\")\n",
    "except Exception as e:\n",
    "    print(\"Error during test set prediction:\", str(e))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['age', 'balance', 'duration', 'campaign', 'pdays', 'previous',\n",
      "       'job_admin.', 'job_blue-collar', 'job_entrepreneur', 'job_housemaid',\n",
      "       'job_management', 'job_retired', 'job_self-employed', 'job_services',\n",
      "       'job_student', 'job_technician', 'job_unemployed', 'marital_married',\n",
      "       'marital_single', 'education_primary', 'education_secondary',\n",
      "       'education_tertiary', 'default_yes', 'housing_yes', 'loan_yes',\n",
      "       'contact_cellular', 'contact_telephone', 'poutcome_failure',\n",
      "       'poutcome_other', 'poutcome_success', 'last_contact_month_2',\n",
      "       'last_contact_month_3', 'last_contact_month_4', 'last_contact_month_5',\n",
      "       'last_contact_month_6', 'last_contact_month_7', 'last_contact_month_8',\n",
      "       'last_contact_month_9', 'last_contact_month_10',\n",
      "       'last_contact_month_11', 'last_contact_month_12'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(X_train.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Numeric features: ['age', 'balance', 'day', 'duration', 'campaign', 'pdays', 'previous']\n",
      "Categorical features: ['job_admin.', 'job_blue-collar', 'job_entrepreneur', 'job_housemaid', 'job_management', 'job_retired', 'job_self-employed', 'job_services', 'job_student', 'job_technician', 'job_unemployed', 'marital_married', 'marital_single', 'education_primary', 'education_secondary', 'education_tertiary', 'default_yes', 'housing_yes', 'loan_yes', 'contact_cellular', 'contact_telephone', 'poutcome_failure', 'poutcome_other', 'poutcome_success', 'last_contact_month_2', 'last_contact_month_3', 'last_contact_month_4', 'last_contact_month_5', 'last_contact_month_6', 'last_contact_month_7', 'last_contact_month_8', 'last_contact_month_9', 'last_contact_month_10', 'last_contact_month_11', 'last_contact_month_12']\n"
     ]
    }
   ],
   "source": [
    "print(\"Numeric features:\", numeric_features)\n",
    "print(\"Categorical features:\", categorical_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Columns in X_train: {'education_primary', 'contact_cellular', 'job_unemployed', 'marital_married', 'previous', 'job_blue-collar', 'default_yes', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'duration', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'campaign', 'last_contact_month_2', 'age', 'pdays', 'balance', 'poutcome_other', 'housing_yes', 'last_contact_month_5', 'last_contact_month_3', 'job_housemaid'}\n",
      "Columns in X_val: {'education_primary', 'contact_cellular', 'job_unemployed', 'marital_married', 'previous', 'job_blue-collar', 'default_yes', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'duration', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'campaign', 'last_contact_month_2', 'age', 'pdays', 'balance', 'poutcome_other', 'housing_yes', 'last_contact_month_5', 'last_contact_month_3', 'job_housemaid'}\n",
      "Columns in df_test_clean: {'education_primary', 'contact_cellular', 'job_unemployed', 'marital_married', 'previous', 'job_blue-collar', 'default_yes', 'last_contact_month_4', 'last_contact_month_9', 'job_self-employed', 'loan_yes', 'job_admin.', 'education_secondary', 'contact_telephone', 'poutcome_failure', 'job_management', 'last_contact_month_7', 'last_contact_month_12', 'job_services', 'duration', 'marital_single', 'job_student', 'last_contact_month_8', 'last_contact_month_11', 'last_contact_month_6', 'poutcome_success', 'education_tertiary', 'job_technician', 'job_retired', 'job_entrepreneur', 'last_contact_month_10', 'campaign', 'last_contact_month_2', 'age', 'pdays', 'balance', 'poutcome_other', 'housing_yes', 'last_contact_month_5', 'last_contact_month_3', 'job_housemaid'}\n"
     ]
    }
   ],
   "source": [
    "df_test = pd.read_csv('test.csv')\n",
    "\n",
    "df_test_clean = clean_data(df_test, is_train=False)\n",
    "\n",
    "print(\"Columns in X_train:\", set(X_train.columns))\n",
    "print(\"Columns in X_val:\", set(X_val.columns))\n",
    "print(\"Columns in df_test_clean:\", set(df_test_clean.columns))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": ".mlp",
   "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": "3.12.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}