Estimated reading time: 66 minutes 
# numpy and pandas for data manipulation
import numpy as np
import pandas as pd 
from sklearn.model_selection import KFold, StratifiedKFold
from sklearn.metrics import roc_auc_score
from lightgbm import LGBMClassifier
from sklearn.preprocessing import LabelEncoder
from sklearn.base import clone
import gc
import pandas as pd
import numpy as np
path = "/home/ubuntu/.kaggle/competitions/home-credit-default-risk/"


params = {
         'boosting_type': 'gbdt',
          'max_depth': -1,
          'objective': 'binary',
          'n_estimators': 3485, 
          'nthread': 5,
          'num_leaves': 39,
          'learning_rate': 0.05,
          'max_bin': 512,
          'subsample_for_bin': 200,
          'subsample': 0.36,
          'subsample_freq': 1,
          'colsample_bytree': 0.98,
          'reg_alpha': 8,
          'reg_lambda': 2,
          'min_split_gain': 0.5,
          'min_child_weight': 1,
          'min_child_samples': 5,
          'scale_pos_weight': 1,
          'num_class': 1,
          'metric': 'auc'}

#Great snippet from https://www.kaggle.com/gemartin/load-data-reduce-memory-usage
def reduce_mem_usage(df):
    """ iterate through all the columns of a dataframe and modify the data type
        to reduce memory usage.        
    """
    start_mem = df.memory_usage().sum() / 1024**2
    print('Memory usage of dataframe is {:.2f} MB'.format(start_mem))
    
    for col in df.columns:
        col_type = df[col].dtype
        
        if col_type != object:
            c_min = df[col].min()
            c_max = df[col].max()
            if str(col_type)[:3] == 'int':
                if c_min > np.iinfo(np.int8).min and c_max < np.iinfo(np.int8).max:
                    df[col] = df[col].astype(np.int8)
                elif c_min > np.iinfo(np.int16).min and c_max < np.iinfo(np.int16).max:
                    df[col] = df[col].astype(np.int16)
                elif c_min > np.iinfo(np.int32).min and c_max < np.iinfo(np.int32).max:
                    df[col] = df[col].astype(np.int32)
                elif c_min > np.iinfo(np.int64).min and c_max < np.iinfo(np.int64).max:
                    df[col] = df[col].astype(np.int64)  
            else:
                if c_min > np.finfo(np.float16).min and c_max < np.finfo(np.float16).max:
                    df[col] = df[col].astype(np.float16)
                elif c_min > np.finfo(np.float32).min and c_max < np.finfo(np.float32).max:
                    df[col] = df[col].astype(np.float32)
                else:
                    df[col] = df[col].astype(np.float64)
        #else:
        #    df[col] = df[col].astype('category')

    end_mem = df.memory_usage().sum() / 1024**2
    print('Memory usage after optimization is: {:.2f} MB'.format(end_mem))
    print('Decreased by {:.1f}%'.format(100 * (start_mem - end_mem) / start_mem))
    
    return df
    
    
    
    

train = reduce_mem_usage(pd.read_csv(path + "application_train.csv"))
test = reduce_mem_usage(pd.read_csv(path + "application_test.csv"))
test.loc[:, "is_test"] = True

alldata = pd.concat([train, test], axis=0)
alldata.loc[:, "is_test"] = alldata.loc[:, "is_test"].fillna(False)

num_cols = alldata.select_dtypes(exclude=["object"]).columns
num_cols = [col for col in num_cols if col not in ["SK_ID_CURR", "is_test", "TARGET"]]

del train, test; gc.collect()


bureau_balance = reduce_mem_usage(pd.read_csv(path + "bureau_balance.csv"))
bureau =  reduce_mem_usage(pd.read_csv(path + 'bureau.csv'))
full_bureau = pd.merge(bureau, bureau_balance, on="SK_ID_BUREAU", how="left")

del bureau_balance, bureau
gc.collect()

dataframes = [
    (
        "previous_application",  
        "SK_ID_PREV", 
        reduce_mem_usage(pd.read_csv(path + 'previous_application.csv'))
    ),
    (
        "bureau", 
        "SK_ID_BUREAU",
        full_bureau
    ),
    (
        "POS_CASH_balance",
        "SK_ID_PREV",
        reduce_mem_usage(pd.read_csv(path + "POS_CASH_balance.csv"))
    ),
    (
        "credit_card_balance",
        "SK_ID_PREV",
        reduce_mem_usage(pd.read_csv(path + "credit_card_balance.csv"))
    ),
    (
        "installments_payments",
        "SK_ID_PREV",
        reduce_mem_usage(pd.read_csv(path + "installments_payments.csv"))
    )
]


for name, key, df in dataframes:
    print("Working on %s..." % name, end="")
    cat_cols = df.select_dtypes(include=["object"]).columns
    
    df = pd.get_dummies(
        df, 
        columns=cat_cols, 
        drop_first=True, 
        dummy_na=True
    )
    
    tmp_df_mean = df.groupby("SK_ID_CURR").agg(["mean", "max"]).drop(key, axis=1)
    tmp_df_mean.columns = ["_".join(col) for col in tmp_df_mean.columns.ravel()]
    #tmp_df_mean.loc[:, "%s_count"%name] = df.loc[:, "SK_ID_CURR"].map(df.groupby('SK_ID_CURR').count()[key])
    
    cols_to_keep = [col for col in tmp_df_mean.columns if col not in alldata.columns]
    alldata = pd.merge(
        alldata, 
        tmp_df_mean[cols_to_keep].reset_index(), 
        on="SK_ID_CURR", 
        how="left"
    )
    del tmp_df_mean
    gc.collect()
    print("done")


for name, key, df in dataframes:
    del df; gc.collect()
    
del dataframes; gc.collect()


categorical_cols = [col for col in alldata.select_dtypes(include=["object"]).columns]

#Mean encoding of categorical variables
for col in categorical_cols:
    means = alldata.loc[~alldata.is_test, :].groupby(col)["TARGET"].mean()
    alldata.loc[:, "%s_MEAN" % col] = alldata.loc[:, col].map(means)
    
    #Missing values is filled with global mean
    alldata.loc[:, "%s_MEAN" % col] = alldata.loc[:, "%s_MEAN" % col].fillna(means.mean())
    
    
    
alldata.loc[:, categorical_cols] = alldata.loc[:, categorical_cols].apply(lambda x: LabelEncoder().fit_transform(x.astype(str)))


pd.options.display.max_columns = None
alldata.head()
AMT_ANNUITY AMT_CREDIT AMT_GOODS_PRICE AMT_INCOME_TOTAL AMT_REQ_CREDIT_BUREAU_DAY AMT_REQ_CREDIT_BUREAU_HOUR AMT_REQ_CREDIT_BUREAU_MON AMT_REQ_CREDIT_BUREAU_QRT AMT_REQ_CREDIT_BUREAU_WEEK AMT_REQ_CREDIT_BUREAU_YEAR APARTMENTS_AVG APARTMENTS_MEDI APARTMENTS_MODE BASEMENTAREA_AVG BASEMENTAREA_MEDI BASEMENTAREA_MODE CNT_CHILDREN CNT_FAM_MEMBERS CODE_GENDER COMMONAREA_AVG COMMONAREA_MEDI COMMONAREA_MODE DAYS_BIRTH DAYS_EMPLOYED DAYS_ID_PUBLISH DAYS_LAST_PHONE_CHANGE DAYS_REGISTRATION DEF_30_CNT_SOCIAL_CIRCLE DEF_60_CNT_SOCIAL_CIRCLE ELEVATORS_AVG ELEVATORS_MEDI ELEVATORS_MODE EMERGENCYSTATE_MODE ENTRANCES_AVG ENTRANCES_MEDI ENTRANCES_MODE EXT_SOURCE_1 EXT_SOURCE_2 EXT_SOURCE_3 FLAG_CONT_MOBILE FLAG_DOCUMENT_10 FLAG_DOCUMENT_11 FLAG_DOCUMENT_12 FLAG_DOCUMENT_13 FLAG_DOCUMENT_14 FLAG_DOCUMENT_15 FLAG_DOCUMENT_16 FLAG_DOCUMENT_17 FLAG_DOCUMENT_18 FLAG_DOCUMENT_19 FLAG_DOCUMENT_2 FLAG_DOCUMENT_20 FLAG_DOCUMENT_21 FLAG_DOCUMENT_3 FLAG_DOCUMENT_4 FLAG_DOCUMENT_5 FLAG_DOCUMENT_6 FLAG_DOCUMENT_7 FLAG_DOCUMENT_8 FLAG_DOCUMENT_9 FLAG_EMAIL FLAG_EMP_PHONE FLAG_MOBIL FLAG_OWN_CAR FLAG_OWN_REALTY FLAG_PHONE FLAG_WORK_PHONE FLOORSMAX_AVG FLOORSMAX_MEDI FLOORSMAX_MODE FLOORSMIN_AVG FLOORSMIN_MEDI FLOORSMIN_MODE FONDKAPREMONT_MODE HOUR_APPR_PROCESS_START HOUSETYPE_MODE LANDAREA_AVG LANDAREA_MEDI LANDAREA_MODE LIVE_CITY_NOT_WORK_CITY LIVE_REGION_NOT_WORK_REGION LIVINGAPARTMENTS_AVG LIVINGAPARTMENTS_MEDI LIVINGAPARTMENTS_MODE LIVINGAREA_AVG LIVINGAREA_MEDI LIVINGAREA_MODE NAME_CONTRACT_TYPE NAME_EDUCATION_TYPE NAME_FAMILY_STATUS NAME_HOUSING_TYPE NAME_INCOME_TYPE NAME_TYPE_SUITE NONLIVINGAPARTMENTS_AVG NONLIVINGAPARTMENTS_MEDI NONLIVINGAPARTMENTS_MODE NONLIVINGAREA_AVG NONLIVINGAREA_MEDI NONLIVINGAREA_MODE OBS_30_CNT_SOCIAL_CIRCLE OBS_60_CNT_SOCIAL_CIRCLE OCCUPATION_TYPE ORGANIZATION_TYPE OWN_CAR_AGE REGION_POPULATION_RELATIVE REGION_RATING_CLIENT REGION_RATING_CLIENT_W_CITY REG_CITY_NOT_LIVE_CITY REG_CITY_NOT_WORK_CITY REG_REGION_NOT_LIVE_REGION REG_REGION_NOT_WORK_REGION SK_ID_CURR TARGET TOTALAREA_MODE WALLSMATERIAL_MODE WEEKDAY_APPR_PROCESS_START YEARS_BEGINEXPLUATATION_AVG YEARS_BEGINEXPLUATATION_MEDI YEARS_BEGINEXPLUATATION_MODE YEARS_BUILD_AVG YEARS_BUILD_MEDI YEARS_BUILD_MODE is_test AMT_ANNUITY_mean AMT_ANNUITY_max AMT_APPLICATION_mean AMT_APPLICATION_max AMT_CREDIT_mean AMT_CREDIT_max AMT_DOWN_PAYMENT_mean AMT_DOWN_PAYMENT_max AMT_GOODS_PRICE_mean AMT_GOODS_PRICE_max HOUR_APPR_PROCESS_START_mean HOUR_APPR_PROCESS_START_max NFLAG_LAST_APPL_IN_DAY_mean NFLAG_LAST_APPL_IN_DAY_max RATE_DOWN_PAYMENT_mean RATE_DOWN_PAYMENT_max RATE_INTEREST_PRIMARY_mean RATE_INTEREST_PRIMARY_max RATE_INTEREST_PRIVILEGED_mean RATE_INTEREST_PRIVILEGED_max DAYS_DECISION_mean DAYS_DECISION_max SELLERPLACE_AREA_mean SELLERPLACE_AREA_max CNT_PAYMENT_mean CNT_PAYMENT_max DAYS_FIRST_DRAWING_mean DAYS_FIRST_DRAWING_max DAYS_FIRST_DUE_mean DAYS_FIRST_DUE_max DAYS_LAST_DUE_1ST_VERSION_mean DAYS_LAST_DUE_1ST_VERSION_max DAYS_LAST_DUE_mean DAYS_LAST_DUE_max DAYS_TERMINATION_mean DAYS_TERMINATION_max NFLAG_INSURED_ON_APPROVAL_mean NFLAG_INSURED_ON_APPROVAL_max NAME_CONTRACT_TYPE_Consumer loans_mean NAME_CONTRACT_TYPE_Consumer loans_max NAME_CONTRACT_TYPE_Revolving loans_mean NAME_CONTRACT_TYPE_Revolving loans_max NAME_CONTRACT_TYPE_XNA_mean NAME_CONTRACT_TYPE_XNA_max NAME_CONTRACT_TYPE_nan_mean NAME_CONTRACT_TYPE_nan_max WEEKDAY_APPR_PROCESS_START_MONDAY_mean WEEKDAY_APPR_PROCESS_START_MONDAY_max WEEKDAY_APPR_PROCESS_START_SATURDAY_mean WEEKDAY_APPR_PROCESS_START_SATURDAY_max WEEKDAY_APPR_PROCESS_START_SUNDAY_mean WEEKDAY_APPR_PROCESS_START_SUNDAY_max WEEKDAY_APPR_PROCESS_START_THURSDAY_mean WEEKDAY_APPR_PROCESS_START_THURSDAY_max WEEKDAY_APPR_PROCESS_START_TUESDAY_mean WEEKDAY_APPR_PROCESS_START_TUESDAY_max WEEKDAY_APPR_PROCESS_START_WEDNESDAY_mean WEEKDAY_APPR_PROCESS_START_WEDNESDAY_max WEEKDAY_APPR_PROCESS_START_nan_mean WEEKDAY_APPR_PROCESS_START_nan_max FLAG_LAST_APPL_PER_CONTRACT_Y_mean FLAG_LAST_APPL_PER_CONTRACT_Y_max FLAG_LAST_APPL_PER_CONTRACT_nan_mean FLAG_LAST_APPL_PER_CONTRACT_nan_max NAME_CASH_LOAN_PURPOSE_Business development_mean NAME_CASH_LOAN_PURPOSE_Business development_max NAME_CASH_LOAN_PURPOSE_Buying a garage_mean NAME_CASH_LOAN_PURPOSE_Buying a garage_max NAME_CASH_LOAN_PURPOSE_Buying a holiday home / land_mean NAME_CASH_LOAN_PURPOSE_Buying a holiday home / land_max NAME_CASH_LOAN_PURPOSE_Buying a home_mean NAME_CASH_LOAN_PURPOSE_Buying a home_max NAME_CASH_LOAN_PURPOSE_Buying a new car_mean NAME_CASH_LOAN_PURPOSE_Buying a new car_max NAME_CASH_LOAN_PURPOSE_Buying a used car_mean NAME_CASH_LOAN_PURPOSE_Buying a used car_max NAME_CASH_LOAN_PURPOSE_Car repairs_mean NAME_CASH_LOAN_PURPOSE_Car repairs_max NAME_CASH_LOAN_PURPOSE_Education_mean NAME_CASH_LOAN_PURPOSE_Education_max NAME_CASH_LOAN_PURPOSE_Everyday expenses_mean NAME_CASH_LOAN_PURPOSE_Everyday expenses_max NAME_CASH_LOAN_PURPOSE_Furniture_mean NAME_CASH_LOAN_PURPOSE_Furniture_max NAME_CASH_LOAN_PURPOSE_Gasification / water supply_mean NAME_CASH_LOAN_PURPOSE_Gasification / water supply_max NAME_CASH_LOAN_PURPOSE_Hobby_mean NAME_CASH_LOAN_PURPOSE_Hobby_max NAME_CASH_LOAN_PURPOSE_Journey_mean NAME_CASH_LOAN_PURPOSE_Journey_max NAME_CASH_LOAN_PURPOSE_Medicine_mean NAME_CASH_LOAN_PURPOSE_Medicine_max NAME_CASH_LOAN_PURPOSE_Money for a third person_mean NAME_CASH_LOAN_PURPOSE_Money for a third person_max NAME_CASH_LOAN_PURPOSE_Other_mean NAME_CASH_LOAN_PURPOSE_Other_max NAME_CASH_LOAN_PURPOSE_Payments on other loans_mean NAME_CASH_LOAN_PURPOSE_Payments on other loans_max NAME_CASH_LOAN_PURPOSE_Purchase of electronic equipment_mean NAME_CASH_LOAN_PURPOSE_Purchase of electronic equipment_max NAME_CASH_LOAN_PURPOSE_Refusal to name the goal_mean NAME_CASH_LOAN_PURPOSE_Refusal to name the goal_max NAME_CASH_LOAN_PURPOSE_Repairs_mean NAME_CASH_LOAN_PURPOSE_Repairs_max NAME_CASH_LOAN_PURPOSE_Urgent needs_mean NAME_CASH_LOAN_PURPOSE_Urgent needs_max NAME_CASH_LOAN_PURPOSE_Wedding / gift / holiday_mean NAME_CASH_LOAN_PURPOSE_Wedding / gift / holiday_max NAME_CASH_LOAN_PURPOSE_XAP_mean NAME_CASH_LOAN_PURPOSE_XAP_max NAME_CASH_LOAN_PURPOSE_XNA_mean NAME_CASH_LOAN_PURPOSE_XNA_max NAME_CASH_LOAN_PURPOSE_nan_mean NAME_CASH_LOAN_PURPOSE_nan_max NAME_CONTRACT_STATUS_Canceled_mean NAME_CONTRACT_STATUS_Canceled_max NAME_CONTRACT_STATUS_Refused_mean NAME_CONTRACT_STATUS_Refused_max NAME_CONTRACT_STATUS_Unused offer_mean NAME_CONTRACT_STATUS_Unused offer_max NAME_CONTRACT_STATUS_nan_mean NAME_CONTRACT_STATUS_nan_max NAME_PAYMENT_TYPE_Cashless from the account of the employer_mean NAME_PAYMENT_TYPE_Cashless from the account of the employer_max NAME_PAYMENT_TYPE_Non-cash from your account_mean NAME_PAYMENT_TYPE_Non-cash from your account_max NAME_PAYMENT_TYPE_XNA_mean NAME_PAYMENT_TYPE_XNA_max NAME_PAYMENT_TYPE_nan_mean NAME_PAYMENT_TYPE_nan_max CODE_REJECT_REASON_HC_mean CODE_REJECT_REASON_HC_max CODE_REJECT_REASON_LIMIT_mean CODE_REJECT_REASON_LIMIT_max CODE_REJECT_REASON_SCO_mean CODE_REJECT_REASON_SCO_max CODE_REJECT_REASON_SCOFR_mean CODE_REJECT_REASON_SCOFR_max CODE_REJECT_REASON_SYSTEM_mean CODE_REJECT_REASON_SYSTEM_max CODE_REJECT_REASON_VERIF_mean CODE_REJECT_REASON_VERIF_max CODE_REJECT_REASON_XAP_mean CODE_REJECT_REASON_XAP_max CODE_REJECT_REASON_XNA_mean CODE_REJECT_REASON_XNA_max CODE_REJECT_REASON_nan_mean CODE_REJECT_REASON_nan_max NAME_TYPE_SUITE_Family_mean NAME_TYPE_SUITE_Family_max NAME_TYPE_SUITE_Group of people_mean NAME_TYPE_SUITE_Group of people_max NAME_TYPE_SUITE_Other_A_mean NAME_TYPE_SUITE_Other_A_max NAME_TYPE_SUITE_Other_B_mean NAME_TYPE_SUITE_Other_B_max NAME_TYPE_SUITE_Spouse, partner_mean NAME_TYPE_SUITE_Spouse, partner_max NAME_TYPE_SUITE_Unaccompanied_mean NAME_TYPE_SUITE_Unaccompanied_max NAME_TYPE_SUITE_nan_mean NAME_TYPE_SUITE_nan_max NAME_CLIENT_TYPE_Refreshed_mean NAME_CLIENT_TYPE_Refreshed_max NAME_CLIENT_TYPE_Repeater_mean NAME_CLIENT_TYPE_Repeater_max NAME_CLIENT_TYPE_XNA_mean NAME_CLIENT_TYPE_XNA_max NAME_CLIENT_TYPE_nan_mean NAME_CLIENT_TYPE_nan_max NAME_GOODS_CATEGORY_Animals_mean NAME_GOODS_CATEGORY_Animals_max NAME_GOODS_CATEGORY_Audio/Video_mean NAME_GOODS_CATEGORY_Audio/Video_max NAME_GOODS_CATEGORY_Auto Accessories_mean NAME_GOODS_CATEGORY_Auto Accessories_max NAME_GOODS_CATEGORY_Clothing and Accessories_mean NAME_GOODS_CATEGORY_Clothing and Accessories_max NAME_GOODS_CATEGORY_Computers_mean NAME_GOODS_CATEGORY_Computers_max NAME_GOODS_CATEGORY_Construction Materials_mean NAME_GOODS_CATEGORY_Construction Materials_max NAME_GOODS_CATEGORY_Consumer Electronics_mean NAME_GOODS_CATEGORY_Consumer Electronics_max NAME_GOODS_CATEGORY_Direct Sales_mean NAME_GOODS_CATEGORY_Direct Sales_max NAME_GOODS_CATEGORY_Education_mean NAME_GOODS_CATEGORY_Education_max NAME_GOODS_CATEGORY_Fitness_mean NAME_GOODS_CATEGORY_Fitness_max NAME_GOODS_CATEGORY_Furniture_mean NAME_GOODS_CATEGORY_Furniture_max NAME_GOODS_CATEGORY_Gardening_mean NAME_GOODS_CATEGORY_Gardening_max NAME_GOODS_CATEGORY_Homewares_mean NAME_GOODS_CATEGORY_Homewares_max NAME_GOODS_CATEGORY_House Construction_mean NAME_GOODS_CATEGORY_House Construction_max NAME_GOODS_CATEGORY_Insurance_mean NAME_GOODS_CATEGORY_Insurance_max NAME_GOODS_CATEGORY_Jewelry_mean NAME_GOODS_CATEGORY_Jewelry_max NAME_GOODS_CATEGORY_Medical Supplies_mean NAME_GOODS_CATEGORY_Medical Supplies_max NAME_GOODS_CATEGORY_Medicine_mean NAME_GOODS_CATEGORY_Medicine_max NAME_GOODS_CATEGORY_Mobile_mean NAME_GOODS_CATEGORY_Mobile_max NAME_GOODS_CATEGORY_Office Appliances_mean NAME_GOODS_CATEGORY_Office Appliances_max NAME_GOODS_CATEGORY_Other_mean NAME_GOODS_CATEGORY_Other_max NAME_GOODS_CATEGORY_Photo / Cinema Equipment_mean NAME_GOODS_CATEGORY_Photo / Cinema Equipment_max NAME_GOODS_CATEGORY_Sport and Leisure_mean NAME_GOODS_CATEGORY_Sport and Leisure_max NAME_GOODS_CATEGORY_Tourism_mean NAME_GOODS_CATEGORY_Tourism_max NAME_GOODS_CATEGORY_Vehicles_mean NAME_GOODS_CATEGORY_Vehicles_max NAME_GOODS_CATEGORY_Weapon_mean NAME_GOODS_CATEGORY_Weapon_max NAME_GOODS_CATEGORY_XNA_mean NAME_GOODS_CATEGORY_XNA_max NAME_GOODS_CATEGORY_nan_mean NAME_GOODS_CATEGORY_nan_max NAME_PORTFOLIO_Cars_mean NAME_PORTFOLIO_Cars_max NAME_PORTFOLIO_Cash_mean NAME_PORTFOLIO_Cash_max NAME_PORTFOLIO_POS_mean NAME_PORTFOLIO_POS_max NAME_PORTFOLIO_XNA_mean NAME_PORTFOLIO_XNA_max NAME_PORTFOLIO_nan_mean NAME_PORTFOLIO_nan_max NAME_PRODUCT_TYPE_walk-in_mean NAME_PRODUCT_TYPE_walk-in_max NAME_PRODUCT_TYPE_x-sell_mean NAME_PRODUCT_TYPE_x-sell_max NAME_PRODUCT_TYPE_nan_mean NAME_PRODUCT_TYPE_nan_max CHANNEL_TYPE_Car dealer_mean CHANNEL_TYPE_Car dealer_max CHANNEL_TYPE_Channel of corporate sales_mean CHANNEL_TYPE_Channel of corporate sales_max CHANNEL_TYPE_Contact center_mean CHANNEL_TYPE_Contact center_max CHANNEL_TYPE_Country-wide_mean CHANNEL_TYPE_Country-wide_max CHANNEL_TYPE_Credit and cash offices_mean CHANNEL_TYPE_Credit and cash offices_max CHANNEL_TYPE_Regional / Local_mean CHANNEL_TYPE_Regional / Local_max CHANNEL_TYPE_Stone_mean CHANNEL_TYPE_Stone_max CHANNEL_TYPE_nan_mean CHANNEL_TYPE_nan_max NAME_SELLER_INDUSTRY_Clothing_mean NAME_SELLER_INDUSTRY_Clothing_max NAME_SELLER_INDUSTRY_Connectivity_mean NAME_SELLER_INDUSTRY_Connectivity_max NAME_SELLER_INDUSTRY_Construction_mean NAME_SELLER_INDUSTRY_Construction_max NAME_SELLER_INDUSTRY_Consumer electronics_mean NAME_SELLER_INDUSTRY_Consumer electronics_max NAME_SELLER_INDUSTRY_Furniture_mean NAME_SELLER_INDUSTRY_Furniture_max NAME_SELLER_INDUSTRY_Industry_mean NAME_SELLER_INDUSTRY_Industry_max NAME_SELLER_INDUSTRY_Jewelry_mean NAME_SELLER_INDUSTRY_Jewelry_max NAME_SELLER_INDUSTRY_MLM partners_mean NAME_SELLER_INDUSTRY_MLM partners_max NAME_SELLER_INDUSTRY_Tourism_mean NAME_SELLER_INDUSTRY_Tourism_max NAME_SELLER_INDUSTRY_XNA_mean NAME_SELLER_INDUSTRY_XNA_max NAME_SELLER_INDUSTRY_nan_mean NAME_SELLER_INDUSTRY_nan_max NAME_YIELD_GROUP_high_mean NAME_YIELD_GROUP_high_max NAME_YIELD_GROUP_low_action_mean NAME_YIELD_GROUP_low_action_max NAME_YIELD_GROUP_low_normal_mean NAME_YIELD_GROUP_low_normal_max NAME_YIELD_GROUP_middle_mean NAME_YIELD_GROUP_middle_max NAME_YIELD_GROUP_nan_mean NAME_YIELD_GROUP_nan_max PRODUCT_COMBINATION_Card X-Sell_mean PRODUCT_COMBINATION_Card X-Sell_max PRODUCT_COMBINATION_Cash_mean PRODUCT_COMBINATION_Cash_max PRODUCT_COMBINATION_Cash Street: high_mean PRODUCT_COMBINATION_Cash Street: high_max PRODUCT_COMBINATION_Cash Street: low_mean PRODUCT_COMBINATION_Cash Street: low_max PRODUCT_COMBINATION_Cash Street: middle_mean PRODUCT_COMBINATION_Cash Street: middle_max PRODUCT_COMBINATION_Cash X-Sell: high_mean PRODUCT_COMBINATION_Cash X-Sell: high_max PRODUCT_COMBINATION_Cash X-Sell: low_mean PRODUCT_COMBINATION_Cash X-Sell: low_max PRODUCT_COMBINATION_Cash X-Sell: middle_mean PRODUCT_COMBINATION_Cash X-Sell: middle_max PRODUCT_COMBINATION_POS household with interest_mean PRODUCT_COMBINATION_POS household with interest_max PRODUCT_COMBINATION_POS household without interest_mean PRODUCT_COMBINATION_POS household without interest_max PRODUCT_COMBINATION_POS industry with interest_mean PRODUCT_COMBINATION_POS industry with interest_max PRODUCT_COMBINATION_POS industry without interest_mean PRODUCT_COMBINATION_POS industry without interest_max PRODUCT_COMBINATION_POS mobile with interest_mean PRODUCT_COMBINATION_POS mobile with interest_max PRODUCT_COMBINATION_POS mobile without interest_mean PRODUCT_COMBINATION_POS mobile without interest_max PRODUCT_COMBINATION_POS other with interest_mean PRODUCT_COMBINATION_POS other with interest_max PRODUCT_COMBINATION_POS others without interest_mean PRODUCT_COMBINATION_POS others without interest_max PRODUCT_COMBINATION_nan_mean PRODUCT_COMBINATION_nan_max DAYS_CREDIT_mean DAYS_CREDIT_max CREDIT_DAY_OVERDUE_mean CREDIT_DAY_OVERDUE_max DAYS_CREDIT_ENDDATE_mean DAYS_CREDIT_ENDDATE_max DAYS_ENDDATE_FACT_mean DAYS_ENDDATE_FACT_max AMT_CREDIT_MAX_OVERDUE_mean AMT_CREDIT_MAX_OVERDUE_max CNT_CREDIT_PROLONG_mean CNT_CREDIT_PROLONG_max AMT_CREDIT_SUM_mean AMT_CREDIT_SUM_max AMT_CREDIT_SUM_DEBT_mean AMT_CREDIT_SUM_DEBT_max AMT_CREDIT_SUM_LIMIT_mean AMT_CREDIT_SUM_LIMIT_max AMT_CREDIT_SUM_OVERDUE_mean AMT_CREDIT_SUM_OVERDUE_max DAYS_CREDIT_UPDATE_mean DAYS_CREDIT_UPDATE_max MONTHS_BALANCE_mean MONTHS_BALANCE_max CREDIT_ACTIVE_Bad debt_mean CREDIT_ACTIVE_Bad debt_max CREDIT_ACTIVE_Closed_mean CREDIT_ACTIVE_Closed_max CREDIT_ACTIVE_Sold_mean CREDIT_ACTIVE_Sold_max CREDIT_ACTIVE_nan_mean CREDIT_ACTIVE_nan_max CREDIT_CURRENCY_currency 2_mean CREDIT_CURRENCY_currency 2_max CREDIT_CURRENCY_currency 3_mean CREDIT_CURRENCY_currency 3_max CREDIT_CURRENCY_currency 4_mean CREDIT_CURRENCY_currency 4_max CREDIT_CURRENCY_nan_mean CREDIT_CURRENCY_nan_max CREDIT_TYPE_Car loan_mean CREDIT_TYPE_Car loan_max CREDIT_TYPE_Cash loan (non-earmarked)_mean CREDIT_TYPE_Cash loan (non-earmarked)_max CREDIT_TYPE_Consumer credit_mean CREDIT_TYPE_Consumer credit_max CREDIT_TYPE_Credit card_mean CREDIT_TYPE_Credit card_max CREDIT_TYPE_Interbank credit_mean CREDIT_TYPE_Interbank credit_max CREDIT_TYPE_Loan for business development_mean CREDIT_TYPE_Loan for business development_max CREDIT_TYPE_Loan for purchase of shares (margin lending)_mean CREDIT_TYPE_Loan for purchase of shares (margin lending)_max CREDIT_TYPE_Loan for the purchase of equipment_mean CREDIT_TYPE_Loan for the purchase of equipment_max CREDIT_TYPE_Loan for working capital replenishment_mean CREDIT_TYPE_Loan for working capital replenishment_max CREDIT_TYPE_Microloan_mean CREDIT_TYPE_Microloan_max CREDIT_TYPE_Mobile operator loan_mean CREDIT_TYPE_Mobile operator loan_max CREDIT_TYPE_Mortgage_mean CREDIT_TYPE_Mortgage_max CREDIT_TYPE_Real estate loan_mean CREDIT_TYPE_Real estate loan_max CREDIT_TYPE_Unknown type of loan_mean CREDIT_TYPE_Unknown type of loan_max CREDIT_TYPE_nan_mean CREDIT_TYPE_nan_max STATUS_1_mean STATUS_1_max STATUS_2_mean STATUS_2_max STATUS_3_mean STATUS_3_max STATUS_4_mean STATUS_4_max STATUS_5_mean STATUS_5_max STATUS_C_mean STATUS_C_max STATUS_X_mean STATUS_X_max STATUS_nan_mean STATUS_nan_max CNT_INSTALMENT_mean CNT_INSTALMENT_max CNT_INSTALMENT_FUTURE_mean CNT_INSTALMENT_FUTURE_max SK_DPD_mean SK_DPD_max SK_DPD_DEF_mean SK_DPD_DEF_max NAME_CONTRACT_STATUS_Amortized debt_mean NAME_CONTRACT_STATUS_Amortized debt_max NAME_CONTRACT_STATUS_Approved_mean NAME_CONTRACT_STATUS_Approved_max NAME_CONTRACT_STATUS_Completed_mean NAME_CONTRACT_STATUS_Completed_max NAME_CONTRACT_STATUS_Demand_mean NAME_CONTRACT_STATUS_Demand_max NAME_CONTRACT_STATUS_Returned to the store_mean NAME_CONTRACT_STATUS_Returned to the store_max NAME_CONTRACT_STATUS_Signed_mean NAME_CONTRACT_STATUS_Signed_max NAME_CONTRACT_STATUS_XNA_mean NAME_CONTRACT_STATUS_XNA_max AMT_BALANCE_mean AMT_BALANCE_max AMT_CREDIT_LIMIT_ACTUAL_mean AMT_CREDIT_LIMIT_ACTUAL_max AMT_DRAWINGS_ATM_CURRENT_mean AMT_DRAWINGS_ATM_CURRENT_max AMT_DRAWINGS_CURRENT_mean AMT_DRAWINGS_CURRENT_max AMT_DRAWINGS_OTHER_CURRENT_mean AMT_DRAWINGS_OTHER_CURRENT_max AMT_DRAWINGS_POS_CURRENT_mean AMT_DRAWINGS_POS_CURRENT_max AMT_INST_MIN_REGULARITY_mean AMT_INST_MIN_REGULARITY_max AMT_PAYMENT_CURRENT_mean AMT_PAYMENT_CURRENT_max AMT_PAYMENT_TOTAL_CURRENT_mean AMT_PAYMENT_TOTAL_CURRENT_max AMT_RECEIVABLE_PRINCIPAL_mean AMT_RECEIVABLE_PRINCIPAL_max AMT_RECIVABLE_mean AMT_RECIVABLE_max AMT_TOTAL_RECEIVABLE_mean AMT_TOTAL_RECEIVABLE_max CNT_DRAWINGS_ATM_CURRENT_mean CNT_DRAWINGS_ATM_CURRENT_max CNT_DRAWINGS_CURRENT_mean CNT_DRAWINGS_CURRENT_max CNT_DRAWINGS_OTHER_CURRENT_mean CNT_DRAWINGS_OTHER_CURRENT_max CNT_DRAWINGS_POS_CURRENT_mean CNT_DRAWINGS_POS_CURRENT_max CNT_INSTALMENT_MATURE_CUM_mean CNT_INSTALMENT_MATURE_CUM_max NAME_CONTRACT_STATUS_Sent proposal_mean NAME_CONTRACT_STATUS_Sent proposal_max NUM_INSTALMENT_VERSION_mean NUM_INSTALMENT_VERSION_max NUM_INSTALMENT_NUMBER_mean NUM_INSTALMENT_NUMBER_max DAYS_INSTALMENT_mean DAYS_INSTALMENT_max DAYS_ENTRY_PAYMENT_mean DAYS_ENTRY_PAYMENT_max AMT_INSTALMENT_mean AMT_INSTALMENT_max AMT_PAYMENT_mean AMT_PAYMENT_max CODE_GENDER_MEAN EMERGENCYSTATE_MODE_MEAN FLAG_OWN_CAR_MEAN FLAG_OWN_REALTY_MEAN FONDKAPREMONT_MODE_MEAN HOUSETYPE_MODE_MEAN NAME_CONTRACT_TYPE_MEAN NAME_EDUCATION_TYPE_MEAN NAME_FAMILY_STATUS_MEAN NAME_HOUSING_TYPE_MEAN NAME_INCOME_TYPE_MEAN NAME_TYPE_SUITE_MEAN OCCUPATION_TYPE_MEAN ORGANIZATION_TYPE_MEAN WALLSMATERIAL_MODE_MEAN WEEKDAY_APPR_PROCESS_START_MEAN
0 24700.5 406597.5 351000.0 202500.0 0.0 0.0 0.0 0.0 0.0 1.0 0.024704 0.024994 0.025208 0.036896 0.036896 0.038300 0 1.0 1 0.014297 0.014397 0.014397 -9461 -637 -2120 -1134.0 -3648.0 2.0 2.0 0.000000 0.000000 0.000000 0 0.068970 0.068970 0.068970 0.083008 0.262939 0.139404 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 1 1 0 0.083313 0.083313 0.083313 0.125000 0.125000 0.125000 3 10 0 0.036896 0.037506 0.037689 0 0 0.020203 0.020493 0.022003 0.018997 0.019302 0.019806 0 4 3 1 7 6 0.000000 0.000000 0.0 0.000000 0.000000 0.0 2.0 2.0 8 5 NaN 0.018799 2 2 0 0 0 0 100002 1.0 0.014900 5 6 0.972168 0.972168 0.972168 0.619141 0.624512 0.634277 False 9251.775391 9251.775391 179055.00 179055.0 179055.00 179055.0 0.000000 0.0 179055.000 179055.0 9.000000 9.0 1.0 1.0 0.000000 0.000000 NaN NaN NaN NaN -606.000000 -606.0 500.000000 500.0 24.000000 24.0 365243.0 365243.0 -565.000000 -565.0 125.000000 125.0 -25.000000 -25.0 -17.000000 -17.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 -996.781818 -103.0 0.0 0.0 -452.75 780.0 -808.5 -36.0 1312.010376 5043.64502 0.0 0.0 111388.835938 450000.000000 70223.140625 245781.0 3198.856445 31988.564453 0.0 0.0 -631.963636 -7.0 -24.554545 0.0 0.0 0.0 0.818182 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.472727 1.0 0.527273 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.245455 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.209091 1.0 0.136364 1.0 0.0 0.0 24.000000 24.0 15.000000 24.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 1.052734 2.0 10.000000 19.0 -295.00 -25.0 -315.5 -49.0 11559.247070 53093.746094 11559.247070 53093.746094 0.101419 0.069649 0.085002 0.079616 0.069782 0.069434 0.083459 0.089399 0.098077 0.077957 0.095885 0.081830 0.105788 0.092996 0.074057 0.081604
1 35698.5 1293502.5 1129500.0 270000.0 0.0 0.0 0.0 0.0 0.0 0.0 0.095886 0.096802 0.092407 0.052887 0.052887 0.053802 0 2.0 0 0.060486 0.060791 0.049713 -16765 -1188 -291 -828.0 -1186.0 0.0 0.0 0.080017 0.080017 0.080627 0 0.034485 0.034485 0.034485 0.311279 0.622070 NaN 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 0 1 0 0.291748 0.291748 0.291748 0.333252 0.333252 0.333252 3 11 0 0.013000 0.013199 0.012802 0 0 0.077271 0.078674 0.078979 0.054901 0.055786 0.055389 0 1 1 1 4 1 0.003901 0.003901 0.0 0.009804 0.010002 0.0 1.0 1.0 3 39 NaN 0.003542 1 1 0 0 0 0 100003 0.0 0.071411 0 1 0.984863 0.984863 0.984863 0.795898 0.798828 0.804199 False 56553.988281 98356.992188 435436.50 900000.0 484191.00 1035882.0 3442.500000 6885.0 435436.500 900000.0 14.666667 17.0 1.0 1.0 0.050018 0.100037 NaN NaN NaN NaN -1305.000000 -746.0 533.000000 1400.0 10.000000 12.0 365243.0 365243.0 -1274.333374 -716.0 -1004.333313 -386.0 -1054.333374 -536.0 -1047.333374 -527.0 0.666504 1.0 0.666667 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.333333 1.0 0.333333 1.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.666667 1.0 0.333333 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.666667 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.000000 0.0 0.666667 1.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.666667 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.333333 1.0 0.000000 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.333333 1.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.333333 1.0 0.666667 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.333333 1.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -1400.750000 -606.0 0.0 0.0 -544.50 1216.0 -1098.0 -540.0 0.000000 0.00000 0.0 0.0 254350.125000 810000.000000 0.000000 0.0 202500.000000 810000.000000 0.0 0.0 -816.000000 -43.0 NaN NaN 0.0 0.0 0.750000 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.500000 1.0 0.500000 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 1.0 1.0 10.109375 12.0 5.785156 12.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.071429 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 1.040039 2.0 5.080000 12.0 -1378.00 -536.0 -1385.0 -544.0 64754.585938 560835.375000 64754.585938 560835.375000 0.069993 0.069649 0.085002 0.083249 0.069782 0.069434 0.083459 0.053551 0.075599 0.077957 0.057550 0.074946 0.063040 0.059148 0.070247 0.077572
2 6750.0 135000.0 135000.0 67500.0 0.0 0.0 0.0 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN 0 1.0 1 NaN NaN NaN -19046 -225 -2531 -815.0 -4260.0 0.0 0.0 NaN NaN NaN 2 NaN NaN NaN NaN 0.556152 0.729492 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 NaN NaN NaN NaN NaN NaN 0 9 1 NaN NaN NaN 0 0 NaN NaN NaN NaN NaN NaN 1 4 3 1 7 6 NaN NaN NaN NaN NaN NaN 0.0 0.0 8 11 26.0 0.010033 2 2 0 0 0 0 100004 0.0 NaN 7 1 NaN NaN NaN NaN NaN NaN False 5357.250000 5357.250000 24282.00 24282.0 20106.00 20106.0 4860.000000 4860.0 24282.000 24282.0 5.000000 5.0 1.0 1.0 0.212036 0.212036 NaN NaN NaN NaN -815.000000 -815.0 30.000000 30.0 4.000000 4.0 365243.0 365243.0 -784.000000 -784.0 -694.000000 -694.0 -724.000000 -724.0 -714.000000 -714.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 1.000000 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.000000 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 -867.000000 -408.0 0.0 0.0 -488.50 -382.0 -532.5 -382.0 0.000000 0.00000 0.0 0.0 94518.898438 94537.796875 0.000000 0.0 0.000000 0.000000 0.0 0.0 -532.000000 -382.0 NaN NaN 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 1.0 1.0 3.750000 4.0 2.250000 4.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.250000 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 1.333008 2.0 2.000000 3.0 -754.00 -724.0 -761.5 -727.0 7096.154785 10573.964844 7096.154785 10573.964844 0.101419 0.082720 0.072437 0.079616 0.067244 0.085273 0.054783 0.089399 0.098077 0.077957 0.095885 0.081830 0.105788 0.069781 0.072915 0.077572
3 29686.5 312682.5 297000.0 135000.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 0 2.0 0 NaN NaN NaN -19005 -3039 -2437 -617.0 -9832.0 0.0 0.0 NaN NaN NaN 2 NaN NaN NaN NaN 0.650391 NaN 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 1 0 0 NaN NaN NaN NaN NaN NaN 0 17 1 NaN NaN NaN 0 0 NaN NaN NaN NaN NaN NaN 0 4 0 1 7 6 NaN NaN NaN NaN NaN NaN 2.0 2.0 8 5 NaN 0.008018 2 2 0 0 0 0 100006 0.0 NaN 7 6 NaN NaN NaN NaN NaN NaN False 23651.175781 39954.511719 272203.25 688500.0 291695.50 906615.0 34840.171875 66987.0 408304.875 688500.0 14.666667 15.0 1.0 1.0 0.163330 0.217773 NaN NaN NaN NaN -272.444444 -181.0 894.222222 8025.0 23.000000 48.0 365243.0 365243.0 91066.500000 365243.0 91584.000000 365243.0 182477.500000 365243.0 182481.750000 365243.0 0.000000 0.0 0.222222 1.0 0.222222 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.111111 1.0 0.111111 1.0 0.666667 1.0 0.111111 1.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.444444 1.0 0.555556 1.0 0.0 0.0 0.333333 1.0 0.111111 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.555556 1.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.888889 1.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.555556 1.0 0.000000 0.0 0.888889 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.777778 1.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.222222 1.0 0.333333 1.0 0.0 0.0 0.000000 0.0 0.444444 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.777778 1.0 0.000000 0.0 0.111111 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.111111 1.0 0.111111 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.777778 1.0 0.0 0.0 0.222222 1.0 0.0 0.0 0.222222 1.0 0.111111 1.0 0.0 0.0 0.111111 1.0 0.222222 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.111111 1.0 0.222222 1.0 0.0 0.0 0.111111 1.0 0.0 0.0 0.111111 1.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 12.000000 48.0 8.648438 48.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.095238 1.0 0.0 0.0 0.047619 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 270000.0 270000.0 NaN NaN 0.0 0.0 NaN NaN NaN NaN 0.0 0.0 NaN NaN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 NaN NaN 0.0 0.0 NaN NaN NaN NaN 0.0 0.0 0.0 0.0 1.125000 2.0 4.437500 10.0 -252.25 -11.0 -271.5 -12.0 62947.085938 691786.875000 62947.085938 691786.875000 0.069993 0.082720 0.085002 0.079616 0.067244 0.085273 0.083459 0.089399 0.099446 0.077957 0.095885 0.081830 0.105788 0.092996 0.072915 0.081604
4 21865.5 513000.0 513000.0 121500.0 0.0 0.0 0.0 0.0 0.0 0.0 NaN NaN NaN NaN NaN NaN 0 1.0 1 NaN NaN NaN -19932 -3038 -3458 -1106.0 -4312.0 0.0 0.0 NaN NaN NaN 2 NaN NaN NaN NaN 0.322754 NaN 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 1 0 1 0 0 NaN NaN NaN NaN NaN NaN 0 11 1 NaN NaN NaN 1 0 NaN NaN NaN NaN NaN NaN 0 4 3 1 7 6 NaN NaN NaN NaN NaN NaN 0.0 0.0 3 37 NaN 0.028656 2 2 0 1 0 0 100007 0.0 NaN 7 4 NaN NaN NaN NaN NaN NaN False 12278.804688 22678.785156 150530.25 247500.0 166638.75 284400.0 3390.750000 3676.5 150530.250 247500.0 12.333333 15.0 1.0 1.0 0.159546 0.218872 NaN NaN NaN NaN -1222.833333 -374.0 409.166667 1200.0 20.671875 48.0 365243.0 365243.0 -1263.199951 -344.0 -837.200012 346.0 72136.203125 365243.0 72143.796875 365243.0 0.600098 1.0 0.333333 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.166667 1.0 0.166667 1.0 0.333333 1.0 0.166667 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.666667 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.166667 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.333333 1.0 0.000000 0.0 0.833333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.666667 1.0 0.0 0.0 0.0 0.0 0.666667 1.0 0.333333 1.0 0.000000 0.0 0.0 0.0 0.166667 1.0 0.500000 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.500000 1.0 0.166667 1.0 0.166667 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.166667 1.0 0.000000 0.0 0.500000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.333333 1.0 0.0 0.0 0.500000 1.0 0.0 0.0 0.000000 0.0 0.500000 1.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.166667 1.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 0.5 1.0 0.166667 1.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.166667 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -1149.000000 -1149.0 0.0 0.0 -783.00 -783.0 -783.0 -783.0 0.000000 0.00000 0.0 0.0 146250.000000 146250.000000 0.000000 0.0 0.000000 0.000000 0.0 0.0 -783.000000 -783.0 NaN NaN 0.0 0.0 1.000000 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.000000 1.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.000000 0.0 0.000000 0.0 1.0 1.0 15.335938 24.0 8.968750 24.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.045455 1.0 0.0 0.0 0.000000 0.0 0.015152 1.0 0.0 0.0 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN 1.166992 2.0 7.045455 17.0 -1029.00 -14.0 -1032.0 -14.0 12666.444336 22678.785156 12214.060547 22678.785156 0.101419 0.082720 0.085002 0.079616 0.067244 0.085273 0.083459 0.089399 0.098077 0.077957 0.095885 0.081830 0.063040 0.058824 0.072915 0.081003 
import pandas as pd 
import numpy as np 

import _pickle as pickle

app_train = pickle.load(open("/data-ml/full_frame_train.p", "rb"))


extra_cols = ['func_cc_CNT_DRAWINGS_ATM_CURRENT',
'func_pos_12_SK_DPD_DEF',
'func_pos_1_AMT_INSTALMENT',
             "SK_ID_CURR"
]

app_train = app_train[extra_cols]

### Start Here

#pickle.dump(alldata, open("/data-ml/alldata.p", "wb"))  # save it into a file 
alldata = pickle.load(open("/data-ml/alldata.p", "rb"))

playdata = alldata.copy()

B_days_columns = ['DAYS_BIRTH', 'DAYS_CREDIT_ENDDATE_mean', 'DAYS_DECISION_mean',
       'DAYS_LAST_PHONE_CHANGE', 'DAYS_ENTRY_PAYMENT_mean',
       'DAYS_LAST_DUE_1ST_VERSION_max', 'DAYS_ENDDATE_FACT_mean',
       'DAYS_CREDIT_mean', 'DAYS_INSTALMENT_max', 'DAYS_ENTRY_PAYMENT_max']

A_days_columns = ['DAYS_ENDDATE_FACT_max', 'DAYS_TERMINATION_mean', 'DAYS_LAST_DUE_max',
       'DAYS_DECISION_max', 'DAYS_EMPLOYED', 'DAYS_CREDIT_ENDDATE_max']

A_mean_columns = ['AMT_ANNUITY_mean', 'RATE_DOWN_PAYMENT_mean', 'DAYS_TERMINATION_mean',
       'NUM_INSTALMENT_VERSION_mean', 'MONTHS_BALANCE_mean',
       'AMT_CREDIT_SUM_LIMIT_mean', 'NAME_YIELD_GROUP_low_action_mean',
       'AMT_CREDIT_SUM_mean', 'PRODUCT_COMBINATION_Cash_mean',
       'CODE_REJECT_REASON_HC_mean', 'CNT_INSTALMENT_FUTURE_mean',
       'STATUS_1_mean', 'CREDIT_ACTIVE_Closed_mean']

B_mean_columns = ['NAME_CONTRACT_STATUS_Refused_mean', 'DAYS_CREDIT_ENDDATE_mean',
       'DAYS_DECISION_mean', 'AMT_CREDIT_mean',
       'CNT_DRAWINGS_CURRENT_mean', 'NAME_CASH_LOAN_PURPOSE_XAP_mean',
       'AMT_PAYMENT_mean', 'WEEKDAY_APPR_PROCESS_START_WEDNESDAY_mean',
       'NAME_PAYMENT_TYPE_XNA_mean',
       'PRODUCT_COMBINATION_Cash X-Sell: low_mean',
       'DAYS_ENTRY_PAYMENT_mean', 'DAYS_ENDDATE_FACT_mean',
       'NAME_YIELD_GROUP_middle_mean', 'DAYS_CREDIT_mean',
       'AMT_INST_MIN_REGULARITY_mean', 'AMT_GOODS_PRICE_mean',
       'NAME_GOODS_CATEGORY_Mobile_mean']

B_max_columns = ['AMT_INSTALMENT_max', 'AMT_DOWN_PAYMENT_max',
       'AMT_CREDIT_MAX_OVERDUE_max', 'RATE_DOWN_PAYMENT_max',
       'AMT_ANNUITY_max', 'DAYS_LAST_DUE_1ST_VERSION_max',
       'DAYS_INSTALMENT_max', 'DAYS_ENTRY_PAYMENT_max']

B_flag_columns = ['FLAG_PHONE', 'FLAG_OWN_CAR', 'FLAG_DOCUMENT_8']

A_flag_columns = ['FLAG_WORK_PHONE', 'NFLAG_INSURED_ON_APPROVAL_mean', 'FLAG_DOCUMENT_3']


playdata["B_DAYS_ALL_Z"] = playdata[B_days_columns].std(axis=1)

playdata["B_Mean_A"] = playdata[B_mean_columns].mean(axis=1)
playdata["B_Mean_Z"] = playdata[B_mean_columns].std(axis=1)

playdata["B_Max_A"] = playdata[B_mean_columns].mean(axis=1)
playdata["B_Max_Z"] = playdata[B_mean_columns].std(axis=1)

#playdata["EXT_SOURCE_All_sum"] = playdata["EXT_SOURCE_1"]  + playdata["EXT_SOURCE_2"] + playdata["EXT_SOURCE_3"] 

#playdata["EXT_SOURCE_All_mult"] = (playdata["EXT_SOURCE_1"]  * playdata["EXT_SOURCE_2"])  / playdata["EXT_SOURCE_3"] 

playdata["an_pay"] = playdata["AMT_ANNUITY"] - playdata["AMT_PAYMENT_mean"]

playdata["inst_pay"] = playdata["AMT_INSTALMENT_mean"] - playdata["AMT_PAYMENT_mean"]

playdata["an_m"] = playdata["AMT_ANNUITY_mean"] - playdata["AMT_PAYMENT_mean"]

playdata["dra_instal"] = playdata["CNT_DRAWINGS_CURRENT_mean"]/playdata["CNT_INSTALMENT_MATURE_CUM_mean"]


playdata["day_diff"] = playdata["DAYS_ENTRY_PAYMENT_mean"] - playdata["DAYS_DECISION_mean"]

playdata["amt_cr"] = playdata["AMT_APPLICATION_mean"] - playdata["AMT_CREDIT_mean"]

playdata["days_w"] = playdata["DAYS_FIRST_DUE_mean"] - playdata["DAYS_LAST_DUE_mean"]

playdata["due_dil"] = playdata["DAYS_LAST_DUE_1ST_VERSION_mean"] - playdata["DAYS_FIRST_DUE_mean"]

playdata["avg_amt"] = playdata["AMT_BALANCE_mean"]/playdata["MONTHS_BALANCE_mean"]


playdata["Mean_Max_ANN"] = playdata['AMT_ANNUITY_max']/playdata['AMT_ANNUITY_mean']

playdata["Mean_Max_AMT"] = playdata['AMT_PAYMENT_max']/playdata['AMT_PAYMENT_mean']

playdata["Mean_Max_CNT"] = playdata['CNT_PAYMENT_max']/playdata['CNT_PAYMENT_mean']

playdata["Mean_Max_RDP"] = playdata['RATE_DOWN_PAYMENT_max']/playdata['RATE_DOWN_PAYMENT_mean']

playdata["Mean_Max_APT"] = playdata['AMT_PAYMENT_TOTAL_CURRENT_max']/playdata['AMT_PAYMENT_TOTAL_CURRENT_mean']

#playdata["EXT_SOURCE_All_sum"] = playdata["EXT_SOURCE_1"]  * playdata["EXT_SOURCE_2"] + playdata["EXT_SOURCE_3"] 

#playdata["EXT_SOURCE_All_mult"] = (playdata["EXT_SOURCE_1"]  * playdata["EXT_SOURCE_2"])  / playdata["EXT_SOURCE_3"] 

playdata.head()
AMT_ANNUITY AMT_CREDIT AMT_GOODS_PRICE AMT_INCOME_TOTAL AMT_REQ_CREDIT_BUREAU_DAY AMT_REQ_CREDIT_BUREAU_HOUR AMT_REQ_CREDIT_BUREAU_MON AMT_REQ_CREDIT_BUREAU_QRT AMT_REQ_CREDIT_BUREAU_WEEK AMT_REQ_CREDIT_BUREAU_YEAR ... day_diff amt_cr days_w due_dil avg_amt Mean_Max_ANN Mean_Max_AMT Mean_Max_CNT Mean_Max_RDP Mean_Max_APT
0 24700.5 406597.5 351000.0 202500.0 0.0 0.0 0.0 0.0 0.0 1.0 ... 290.500000 0.00 -540.00000 690.000000 NaN 1.000000 4.593184 1.000000 NaN NaN
1 35698.5 1293502.5 1129500.0 270000.0 0.0 0.0 0.0 0.0 0.0 0.0 ... -80.000000 -48754.50 -220.00000 270.000061 NaN 1.739170 8.660936 1.200195 2.000000 NaN
2 6750.0 135000.0 135000.0 67500.0 0.0 0.0 0.0 0.0 0.0 0.0 ... 53.500000 4176.00 -60.00000 90.000000 NaN 1.000000 1.490098 1.000000 1.000000 NaN
3 29686.5 312682.5 297000.0 135000.0 NaN NaN NaN NaN NaN NaN ... 0.944444 -19492.25 -91411.00000 517.500000 NaN 1.689324 10.989975 2.087891 1.333008 NaN
4 21865.5 513000.0 513000.0 121500.0 0.0 0.0 0.0 0.0 0.0 0.0 ... 190.833333 -16108.50 -73399.40625 425.999939 NaN 1.846986 1.856777 2.322266 1.372070 NaN

5 rows × 638 columns

##playdata= pd.merge(playdata, principalDf, on="SK_ID_CURR", how="left")
playdata= pd.merge(playdata, app_train, on="SK_ID_CURR", how="left")

playdata = playdata.replace([np.inf, -np.inf], np.nan)

playdata = playdata.fillna(value=0)
print(playdata.shape)
playdata = playdata.dropna(axis=1)
print(playdata.shape)

(356255, 641)
(356255, 641)

genetic_columns = ['i136', 'i12', 'i10', 'i17', 'i8', 'i13', 'i35', 'i24', 'i14', 'i9',
       'i2', 'i119', 'i15', 'i4', 'i11', 'i21', 'i127', 'i20', 'i19',
       'i32', 'i22', 'i28', 'i26', 'i317', 'i37', 'i3', 'i1', 'i51', 'i43',
       'i23', 'i18', 'i126', 'i105', 'i27', 'i494', 'i16', 'i34', 'i404',
       'i36', 'i454', 'i62', 'i55', 'i340', 'i25', 'i413', 'i7', 'i39',
       'i304', 'i111', 'i74', 'i382', 'i67', 'i57', 'i68', 'i6', 'i348',
       'i175', 'i493', 'i289', 'i171', 'i30', 'i64', 'i255', 'i94', 'i290',
       'i50', 'i140', 'i176', 'i33', 'i491', 'i5', 'i70', 'i489', 'i38',
       'i29', 'i45', 'i152', 'i63', 'i205', 'i69', 'i325', 'i374', 'i326',
       'i101', 'i77', 'i227', 'i390', 'i251', 'i183', 'i244', 'i149',
       'i445', 'i490', 'i0', 'i381', 'i300', 'i228', 'i403', 'i224',
       'i311', 'i359', 'i177', 'i192', 'i173', 'i298', 'i234', 'i40',
       'i129', 'i419', 'i318', 'i384', 'i236', 'i378', 'i258', 'i47',
       'i405', 'i502', 'i492', 'i411', 'i53', 'i448', 'i343', 'i389',
       'i211', 'i60', 'i420', 'i31', 'i391', 'i66', 'i363', 'i108', 'i379',
       'i130', 'i406', 'i280', 'i498', 'i172', 'i483', 'i85', 'i237',
       'i260', 'i131', 'i431', 'i464', 'i100', 'i444', 'i257', 'i504',
       'i87', 'i435', 'i294', 'i386', 'i48', 'i150', 'i242', 'i505',
       'i156', 'i426', 'i256', 'i46', 'i49', 'i123', 'i501', 'i352', 'i56',
       'i125', 'i249', 'i380', 'i114', 'i75', 'i440', 'i99', 'i485',
       'i190', 'i477', 'i309', 'i424', 'i214', 'i407', 'i254', 'i135',
       'i218', 'i368', 'i73', 'i253', 'i117', 'i98', 'i216', 'i230',
       'i328', 'i361', 'i484', 'i365', 'i315', 'i96', 'i265', 'i148',
       'i122', 'i181', 'i417', 'i295', 'i503', 'i385', 'i42', 'i162',
       'i147', 'i185', 'i79', 'i141', 'i157', 'i213', 'i450', 'i59', 'i84',
       'i118', 'i203', 'i110', 'i120', 'i495', 'i347', 'i144', 'i324',
       'i128', 'i331', 'i303', 'i138', 'i235', 'i262', 'i170', 'i240',
       'i232', 'i344', 'i198', 'i54', 'i137', 'i451', 'i456', 'i212',
       'i233', 'i394', 'i443', 'i310', 'i442', 'i356', 'i158', 'i160',
       'i102', 'i179', 'i103', 'i65', 'i246', 'i510', 'i239', 'i428',
       'i460', 'i285', 'i97', 'i387', 'i480', 'i146', 'i319', 'i113',
       'i291', 'i91', 'i95', 'i169', 'i82', 'i89', 'i193', 'i468', 'i409',
       'i52', 'i376', 'i115', 'i116', 'i400', 'i151', 'i441', 'i266',
       'i399', 'i307', 'i44', 'i112', 'i305', 'i72', 'i459', 'i342',
       'i458', 'i182', 'i471', 'i372', 'i507', 'i221', 'i369', 'i167',
       'i455', 'i299', 'i139', 'i377', 'i497', 'SK_ID_CURR']

genetic_test = pickle.load(open("/data-ml/genetic_test.p", "rb"))
genetic_train = pickle.load(open("/data-ml/genetic_train.p", "rb"))

genetic_train = pd.concat((genetic_train[genetic_columns],genetic_test[genetic_columns]),axis=0)

del genetic_test

playdata = pd.merge(playdata,genetic_train, on="SK_ID_CURR",how="left" )

del genetic_train

1+1

2

Each time-step takes 30 seconds.

function_set = [‘add’, ‘sub’, ‘mul’, ‘div’, ‘sqrt’, ‘log’, ‘abs’, ‘neg’, ‘inv’,’tan’]

gp = SymbolicTransformer(generations=800, population_size=200, hall_of_fame=100, n_components=10, function_set=function_set, parsimony_coefficient=0.0005, max_samples=0.9, verbose=1, random_state=0, n_jobs=6)

gp.fit(playdata.loc[~playdata.is_test, :].drop([“SK_ID_CURR”, “TARGET”, “is_test”], axis=1), playdata.loc[~playdata.is_test, :][“TARGET”]) gp_features.to_csv(“gp_features.csv”,index=False)

new = pd.read_csv("imp_feats2.csv")

fit_list  = list(new.iloc[:,0].values)

playdata = playdata.drop(fit_list,axis=1)

playdata.shape

(356255, 435)

import _pickle as pickle
#pickle.dump(playdata, open("/data-ml/playdata.p", "wb"))  # save it into a file 
playdata = pickle.load(open("/data-ml/playdata.p", "rb"))

low = ['i7', 'i8', 'i9', 'i10', 'i11']

high = ['i3', 'i15', 'i12', 'i35', 'EXT_SOURCE_3']

low_l = ['i11', 'i10', 'i9', 'i8', 'i7']

high_l = ['i1', 'inst_pay', 'DAYS_CREDIT_UPDATE_mean', 'i25', 'i13']

for h, l in zip(high, low):
    playdata[h+"_"+l] = playdata[l]/playdata[h]
    playdata[h+"_"+l] = playdata[h+"_"+l].replace([np.inf, -np.inf], np.nan)
    playdata[h+"_"+l] = playdata[h+"_"+l].fillna(value=0)

#playdata = playdata.reset_index()

smalldata = playdata.loc[~playdata.is_test, :].iloc[140000:200000,:]

cols_to_drop = ["SK_ID_CURR", "TARGET", "is_test"]
## Can possibly include, SK_ID_CURR

X_train = smalldata.loc[~smalldata.is_test, :].drop(cols_to_drop, axis=1)

y_train = smalldata.loc[~smalldata.is_test, "TARGET"]

LGB and XGB have a rich toolset to remove noisy features and regularize your models. Two of the most important for this competition are feature_fraction and reg_lambda.

# X_train = alldata.loc[~alldata.is_test, :].drop(cols_to_drop, axis=1)
# y_train = alldata.loc[~alldata.is_test, "TARGET"]

X_test = playdata.loc[playdata.is_test, :].drop(cols_to_drop, axis=1)


n_splits = 5
cvv = StratifiedKFold(n_splits=n_splits, random_state=42)

oof_preds = np.zeros(X_train.shape[0])

sub = pd.read_csv(path + "sample_submission.csv")
sub["TARGET"] = 0
feature_importances = pd.DataFrame()

for i, (fit_idx, val_idx) in enumerate(cvv.split(X_train, y_train)):
    
    X_fit = X_train.iloc[fit_idx]
    y_fit = y_train.iloc[fit_idx]
    X_val = X_train.iloc[val_idx]
    y_val = y_train.iloc[val_idx]
    
    model = LGBMClassifier(
**params
    )

    model.fit(
        X_fit,
        y_fit,
        eval_set=[(X_fit, y_fit), (X_val, y_val)],
        eval_names=('fit', 'val'),
        eval_metric='auc',
        early_stopping_rounds=150,
        verbose=False
    )
    
    
    oof_preds[val_idx] = model.predict_proba(X_val, num_iteration=model.best_iteration_)[:, 1]
    sub['TARGET'] += model.predict_proba(X_test, num_iteration=model.best_iteration_)[:,1]
    
    fi = pd.DataFrame()
    fi["feature"] = X_train.columns
    fi["importance"] = model.feature_importances_
    fi["fold"] = (i+1)
    
    feature_importances = pd.concat([feature_importances, fi], axis=0)
    
    print("Fold {} AUC: {:.8f}".format(i+1, roc_auc_score(y_val, oof_preds[val_idx])))
    
print('Full AUC score %.8f' % roc_auc_score(y_train, oof_preds))  

Fold 1 AUC: 0.77929129
Fold 2 AUC: 0.77723532
Fold 3 AUC: 0.77136001
Fold 4 AUC: 0.77641178
Fold 5 AUC: 0.77529514
Full AUC score 0.77585398

Fold 1 AUC: 0.77349891
Fold 2 AUC: 0.77088397
Fold 3 AUC: 0.77019909
Fold 4 AUC: 0.77242054
Fold 5 AUC: 0.77165830
Full AUC score 0.77165772

params = {
         'boosting_type': 'gbdt',
          'max_depth': -1,
          'objective': 'binary',
          'n_estimators': 3485, 
          'nthread': 5,
          'num_leaves': 39,
          'learning_rate': 0.05,
          'max_bin': 512,
          'subsample_for_bin': 200,
          'subsample': 0.36,
          'subsample_freq': 1,
          'colsample_bytree': 0.98,
          'reg_alpha': 8,
          'reg_lambda': 2,
          'min_split_gain': 0.5,
          'min_child_weight': 1,
          'min_child_samples': 5,
          'scale_pos_weight': 1,
          'num_class': 1,
          'metric': 'auc'}

## Someones

params["max_depth"]=5
params["num_leaves"]=5 ** 2 - 1
params["learning_rate"]=0.007
params["n_estimators"]=30000
params["min_child_samples"]=80
params["subsample"]=0.8
params["colsample_bytree"]=1

# params["reg_lambda"] = 2
# params["feature_fraction and"] = 2
# params["seed"] = 5

#### Scripus/First Olivier

params['n_estimators']=4000
params['learning_rate']=0.03
params['num_leaves']=30
params['colsample_bytree']=.8

params['subsample']=.9
params['max_depth']=7

params['reg_alpha']=.1

params['reg_lambda']=.1
params['min_split_gain']=.01
params['min_child_weight']=100
params['silent']=-1
params['verbose']=-1
params['nthread']=4

## More Overfitting.
## Better First Outcome - Worth Trying

params['max_depth']=5
params['reg_alpha']=.32
params['colsample_bytree']=.70
params['learning_rate']=0.025

#### Aguiar/Bayes Opt
#### Whatever benefit you got out of first olivier - this would give you 0.001 more. 


params['n_estimators']=10000
params['learning_rate']=0.02
params['num_leaves']=34
params['colsample_bytree']=0.9497036
params['subsample']=0.8715623
params['max_depth']=8
params['reg_alpha']=0.041545473
params['reg_lambda']=0.0735294
params['min_split_gain']=0.0222415
params['min_child_weight']=39.3259775
params['silent']=-1
params['verbose']=-1


from lightgbm import cv
from lightgbm import Dataset
def get_score(X, y, usecols, params, dropcols=[]):  
     dtrain = Dataset(X[usecols].drop(dropcols, axis=1), y)
     eval =  cv(params,
             dtrain,
             nfold=5,
             stratified=True,
             num_boost_round=20000,
             early_stopping_rounds=320, ## After it stopped how long should go on. 
             verbose_eval=20,
             seed = 5,
             show_stdv=True)
     return max(eval['auc-mean'])

## 1940

get_score(X_train,y_train , list(X_train.columns), params )

/home/ubuntu/anaconda3/lib/python3.6/site-packages/lightgbm/engine.py:390: UserWarning: Found `n_estimators` in params. Will use it instead of argument
  warnings.warn("Found `{}` in params. Will use it instead of argument".format(alias))
/home/ubuntu/anaconda3/lib/python3.6/site-packages/lightgbm/basic.py:657: UserWarning: silent keyword has been found in `params` and will be ignored. Please use silent argument of the Dataset constructor to pass this parameter.
  'Please use {0} argument of the Dataset constructor to pass this parameter.'.format(key))


[20]	cv_agg's auc: 0.750899 + 0.00239836
[40]	cv_agg's auc: 0.753287 + 0.00246512
[60]	cv_agg's auc: 0.755524 + 0.00245156
[80]	cv_agg's auc: 0.758147 + 0.00248164
[100]	cv_agg's auc: 0.760763 + 0.00240347
[120]	cv_agg's auc: 0.763728 + 0.00230278
[140]	cv_agg's auc: 0.766515 + 0.00218787
[160]	cv_agg's auc: 0.769025 + 0.00202398
[180]	cv_agg's auc: 0.77138 + 0.00187914
[200]	cv_agg's auc: 0.773459 + 0.00173328
[220]	cv_agg's auc: 0.775105 + 0.00167595
[240]	cv_agg's auc: 0.776582 + 0.00159076
[260]	cv_agg's auc: 0.777786 + 0.0015023
[280]	cv_agg's auc: 0.778834 + 0.00143714
[300]	cv_agg's auc: 0.779782 + 0.00141863
[320]	cv_agg's auc: 0.780557 + 0.00143464
[340]	cv_agg's auc: 0.78123 + 0.00133199
[360]	cv_agg's auc: 0.781897 + 0.00139311
[380]	cv_agg's auc: 0.782541 + 0.00132599
[400]	cv_agg's auc: 0.783049 + 0.00131854
[420]	cv_agg's auc: 0.783565 + 0.00136944
[440]	cv_agg's auc: 0.783998 + 0.00139678
[460]	cv_agg's auc: 0.784402 + 0.00136716
[480]	cv_agg's auc: 0.784776 + 0.00131997
[500]	cv_agg's auc: 0.785092 + 0.00130671
[520]	cv_agg's auc: 0.785412 + 0.00129445
[540]	cv_agg's auc: 0.785687 + 0.001333
[560]	cv_agg's auc: 0.785975 + 0.00134971
[580]	cv_agg's auc: 0.786204 + 0.00129318
[600]	cv_agg's auc: 0.786451 + 0.0012906
[620]	cv_agg's auc: 0.786658 + 0.00131939
[640]	cv_agg's auc: 0.786836 + 0.00130499
[660]	cv_agg's auc: 0.787018 + 0.00130422
[680]	cv_agg's auc: 0.787212 + 0.00129278
[700]	cv_agg's auc: 0.787389 + 0.00130515
[720]	cv_agg's auc: 0.787528 + 0.00134682
[740]	cv_agg's auc: 0.78765 + 0.00134197
[760]	cv_agg's auc: 0.787781 + 0.00136115
[780]	cv_agg's auc: 0.787922 + 0.00134426
[800]	cv_agg's auc: 0.788075 + 0.00133484
[820]	cv_agg's auc: 0.788184 + 0.00133475
[840]	cv_agg's auc: 0.788312 + 0.00134592
[860]	cv_agg's auc: 0.788427 + 0.00133218
[880]	cv_agg's auc: 0.788538 + 0.00136188
[900]	cv_agg's auc: 0.788682 + 0.00133944
[920]	cv_agg's auc: 0.788785 + 0.00134624
[940]	cv_agg's auc: 0.788852 + 0.00132261
[960]	cv_agg's auc: 0.788956 + 0.00132642
[980]	cv_agg's auc: 0.789066 + 0.00134892
[1000]	cv_agg's auc: 0.789159 + 0.00133023
[1020]	cv_agg's auc: 0.789224 + 0.00135108
[1040]	cv_agg's auc: 0.789291 + 0.00136096
[1060]	cv_agg's auc: 0.789362 + 0.00138416
[1080]	cv_agg's auc: 0.789473 + 0.0013763
[1100]	cv_agg's auc: 0.789529 + 0.00134436
[1120]	cv_agg's auc: 0.789628 + 0.00135566
[1140]	cv_agg's auc: 0.78969 + 0.00135585
[1160]	cv_agg's auc: 0.789747 + 0.00135354
[1180]	cv_agg's auc: 0.789794 + 0.00137751
[1200]	cv_agg's auc: 0.789855 + 0.00139769
[1220]	cv_agg's auc: 0.789892 + 0.00141634
[1240]	cv_agg's auc: 0.789939 + 0.00146886
[1260]	cv_agg's auc: 0.789978 + 0.00149474
[1280]	cv_agg's auc: 0.790038 + 0.00149766
[1300]	cv_agg's auc: 0.790075 + 0.00147383
[1320]	cv_agg's auc: 0.790096 + 0.00148093
[1340]	cv_agg's auc: 0.79014 + 0.00149847
[1360]	cv_agg's auc: 0.790174 + 0.00152227
[1380]	cv_agg's auc: 0.790238 + 0.00152361
[1400]	cv_agg's auc: 0.790278 + 0.00155701
[1420]	cv_agg's auc: 0.790303 + 0.00155319
[1440]	cv_agg's auc: 0.790331 + 0.00152657
[1460]	cv_agg's auc: 0.79037 + 0.0015125
[1480]	cv_agg's auc: 0.790392 + 0.00151134
[1500]	cv_agg's auc: 0.790396 + 0.00150945
[1520]	cv_agg's auc: 0.790422 + 0.00151041
[1540]	cv_agg's auc: 0.790465 + 0.00150177
[1560]	cv_agg's auc: 0.790507 + 0.00153057
[1580]	cv_agg's auc: 0.790542 + 0.00150971
[1600]	cv_agg's auc: 0.790565 + 0.00151222
[1620]	cv_agg's auc: 0.790608 + 0.00151308
[1640]	cv_agg's auc: 0.790608 + 0.00150761
[1660]	cv_agg's auc: 0.790647 + 0.00149686
[1680]	cv_agg's auc: 0.790658 + 0.00148574
[1700]	cv_agg's auc: 0.790661 + 0.00148581
[1720]	cv_agg's auc: 0.790682 + 0.0014751
[1740]	cv_agg's auc: 0.790665 + 0.00147565
[1760]	cv_agg's auc: 0.790666 + 0.00148164
[1780]	cv_agg's auc: 0.790657 + 0.00148223
[1800]	cv_agg's auc: 0.790643 + 0.00148393
[1820]	cv_agg's auc: 0.790659 + 0.00145436
[1840]	cv_agg's auc: 0.790631 + 0.00143657
[1860]	cv_agg's auc: 0.790619 + 0.001454
[1880]	cv_agg's auc: 0.790609 + 0.00143859
[1900]	cv_agg's auc: 0.790643 + 0.00146536
[1920]	cv_agg's auc: 0.790673 + 0.00146466
[1940]	cv_agg's auc: 0.790673 + 0.00145731
[1960]	cv_agg's auc: 0.790676 + 0.00145561
[1980]	cv_agg's auc: 0.790651 + 0.00142868
[2000]	cv_agg's auc: 0.790638 + 0.00141753
[2020]	cv_agg's auc: 0.79065 + 0.00144084
[2040]	cv_agg's auc: 0.790655 + 0.00140762





0.79070065853297

[20]	cv_agg's auc: 0.745845 + 0.00631848
[40]	cv_agg's auc: 0.748735 + 0.00605509
[60]	cv_agg's auc: 0.751374 + 0.00564571
[80]	cv_agg's auc: 0.753998 + 0.00572914
[100]	cv_agg's auc: 0.756195 + 0.00530572

## Don't Delete
##get_score(X_train,y_train , list(X_train.columns), params )

/home/ubuntu/anaconda3/lib/python3.6/site-packages/lightgbm/engine.py:390: UserWarning: Found `n_estimators` in params. Will use it instead of argument
  warnings.warn("Found `{}` in params. Will use it instead of argument".format(alias))
/home/ubuntu/anaconda3/lib/python3.6/site-packages/lightgbm/basic.py:657: UserWarning: silent keyword has been found in `params` and will be ignored. Please use silent argument of the Dataset constructor to pass this parameter.
  'Please use {0} argument of the Dataset constructor to pass this parameter.'.format(key))


[20]	cv_agg's auc: 0.747892 + 0.00290124
[40]	cv_agg's auc: 0.751845 + 0.00268741
[60]	cv_agg's auc: 0.755811 + 0.00260676
[80]	cv_agg's auc: 0.759646 + 0.00256537
[100]	cv_agg's auc: 0.763551 + 0.00244864
[120]	cv_agg's auc: 0.767361 + 0.00221175
[140]	cv_agg's auc: 0.770545 + 0.00208467
[160]	cv_agg's auc: 0.773334 + 0.00189316
[180]	cv_agg's auc: 0.775621 + 0.00174513
[200]	cv_agg's auc: 0.77762 + 0.00167602
[220]	cv_agg's auc: 0.779307 + 0.00164218
[240]	cv_agg's auc: 0.780757 + 0.00158985
[260]	cv_agg's auc: 0.781872 + 0.00158269
[280]	cv_agg's auc: 0.782887 + 0.00155406
[300]	cv_agg's auc: 0.783723 + 0.0015111
[320]	cv_agg's auc: 0.784442 + 0.00150647
[340]	cv_agg's auc: 0.785076 + 0.00152412
[360]	cv_agg's auc: 0.785549 + 0.00160498
[380]	cv_agg's auc: 0.786153 + 0.00158198
[400]	cv_agg's auc: 0.7865 + 0.0015672
[420]	cv_agg's auc: 0.786869 + 0.00152609
[440]	cv_agg's auc: 0.78717 + 0.00150457
[460]	cv_agg's auc: 0.787452 + 0.00153412
[480]	cv_agg's auc: 0.787748 + 0.00158983
[500]	cv_agg's auc: 0.788004 + 0.00157703
[520]	cv_agg's auc: 0.788219 + 0.00159905
[540]	cv_agg's auc: 0.78852 + 0.00159311
[560]	cv_agg's auc: 0.788696 + 0.00157586
[580]	cv_agg's auc: 0.788886 + 0.00158338
[600]	cv_agg's auc: 0.789052 + 0.00151509
[620]	cv_agg's auc: 0.78921 + 0.00145105
[640]	cv_agg's auc: 0.789319 + 0.00147611
[660]	cv_agg's auc: 0.789408 + 0.00146584
[680]	cv_agg's auc: 0.789553 + 0.00147051
[700]	cv_agg's auc: 0.789634 + 0.00145347
[720]	cv_agg's auc: 0.789737 + 0.00147721
[740]	cv_agg's auc: 0.789837 + 0.00146458
[760]	cv_agg's auc: 0.789945 + 0.00147083
[780]	cv_agg's auc: 0.790051 + 0.00148616
[800]	cv_agg's auc: 0.790099 + 0.00152455
[820]	cv_agg's auc: 0.790166 + 0.00156027
[840]	cv_agg's auc: 0.79021 + 0.00160685
[860]	cv_agg's auc: 0.790269 + 0.00156309
[880]	cv_agg's auc: 0.790303 + 0.00159579
[900]	cv_agg's auc: 0.790334 + 0.00159927
[920]	cv_agg's auc: 0.790381 + 0.00160419
[940]	cv_agg's auc: 0.790369 + 0.00162052
[960]	cv_agg's auc: 0.790383 + 0.00162689
[980]	cv_agg's auc: 0.790423 + 0.00162551
[1000]	cv_agg's auc: 0.790447 + 0.00164424
[1020]	cv_agg's auc: 0.790453 + 0.00164684
[1040]	cv_agg's auc: 0.790481 + 0.00165162
[1060]	cv_agg's auc: 0.790516 + 0.00165075
[1080]	cv_agg's auc: 0.790516 + 0.00163296
[1100]	cv_agg's auc: 0.790523 + 0.00163513
[1120]	cv_agg's auc: 0.790497 + 0.00161327
[1140]	cv_agg's auc: 0.790528 + 0.00160904
[1160]	cv_agg's auc: 0.790539 + 0.00159895
[1180]	cv_agg's auc: 0.790545 + 0.00159921
[1200]	cv_agg's auc: 0.790552 + 0.00160067
[1220]	cv_agg's auc: 0.790563 + 0.00158254
[1240]	cv_agg's auc: 0.790579 + 0.00155828
[1260]	cv_agg's auc: 0.790569 + 0.0015838
[1280]	cv_agg's auc: 0.790564 + 0.00159268
[1300]	cv_agg's auc: 0.790589 + 0.00159251
[1320]	cv_agg's auc: 0.790573 + 0.00160981
[1340]	cv_agg's auc: 0.790579 + 0.00163903
[1360]	cv_agg's auc: 0.790565 + 0.0016914
[1380]	cv_agg's auc: 0.790511 + 0.00164031
[1400]	cv_agg's auc: 0.790487 + 0.0016701
[1420]	cv_agg's auc: 0.790467 + 0.00166841
[1440]	cv_agg's auc: 0.79045 + 0.00165862
[1460]	cv_agg's auc: 0.790436 + 0.00167787
[1480]	cv_agg's auc: 0.790454 + 0.00168202





0.79058929749666795

### brand_new
                                            1580 seed 13             -7902 - Doesn't do well.
### new
###                                          1180 seed 13,  1180 seed 20
### 440, seed 5  380, seed 1 . 380 seed 9 460 seed  13, 420 seed 20  -0.790589 

# dip  = {}
# dip[1] = [440, 5]
# dip[2] = [380, 1]
# dip[3] = [380, 9]
# dip[4] = [460, 13]
# dip[5] = [420, 20]

#model.best_iteration_

dip  = {}
dip[1] = [1180, 5]
dip[2] = [1180, 1]
dip[3] = [1180, 9]
dip[4] = [1180, 13]
dip[5] = [1180, 20]

#model.best_iteration_

# X_train = alldata.loc[~alldata.is_test, :].drop(cols_to_drop, axis=1)
# y_train = alldata.loc[~alldata.is_test, "TARGET"]

X_test = playdata.loc[playdata.is_test, :].drop(cols_to_drop, axis=1)


n_splits = 5
cvv = StratifiedKFold(n_splits=n_splits, random_state=42)

oof_preds = np.zeros(X_train.shape[0])

sub = pd.read_csv(path + "sample_submission.csv")
sub["TARGET"] = 0
feature_importances = pd.DataFrame()

ba= 0 
for i, (fit_idx, val_idx) in enumerate(cvv.split(X_train, y_train)):
    ba = ba + 1
    
    params["seed"] = dip[ba][1]
    params["num_boost_round"] = dip[ba][0]
    model = LGBMClassifier(
**params
    )

    model.fit(
        X_train,
        y_train,
        eval_metric='auc',
        verbose=False
    )
    
    sub['TARGET'] += model.predict_proba(X_test, num_iteration=dip[ba][0])[:,1]
    
    fi = pd.DataFrame()
    fi["feature"] = X_train.columns
    fi["importance"] = model.feature_importances_
    fi["fold"] = (i+1)
    
    feature_importances = pd.concat([feature_importances, fi], axis=0)
    

sub["TARGET"] /= n_splits
sub.to_csv("lgbm6.csv", index=None, float_format="%.8f")

import shap

shap_values = shap.TreeExplainer(model).shap_values(X_train)

shap_fram = pd.DataFrame(shap_values[:,:-1], columns=list(X_train.columns))

shap_new = shap_fram.sum().sort_values().to_frame()

shap_new.columns = ["SHAP"]

list(shap_new.head(20).index.values)

[‘i3’, ‘i15’, ‘i12’, ‘i35’, ‘EXT_SOURCE_3’, ‘i13’, ‘i25’, ‘DAYS_CREDIT_UPDATE_mean’, ‘inst_pay’, ‘i1’, ‘DAYS_CREDIT_mean’, ‘i37’, ‘i77’, ‘AMT_ANNUITY’, ‘i6’, ‘i119’, ‘i18’, ‘i384’, ‘i183’, ‘DAYS_CREDIT_max’]

high = list(shap_new.head(5).index.values)
low = [rev for rev in reversed(list(shap_new.tail(5).index.values))]

low = ['i7', 'i8', 'i9', 'i10', 'i11']

high = ['i3', 'i15', 'i12', 'i35', 'EXT_SOURCE_3']

[‘i7’, ‘i8’, ‘i9’, ‘i10’, ‘i11’]

## The Other Five

[rev for rev in reversed(list(shap_new.head(10).index.values))][:5]

[‘i1’, ‘inst_pay’, ‘DAYS_CREDIT_UPDATE_mean’, ‘i25’, ‘i13’]

list(shap_new.tail(5).index.values)

[‘i11’, ‘i10’, ‘i9’, ‘i8’, ‘i7’]

shap_new.head()
SHAP
i3 -362.615099
i15 -67.921697
i12 -59.016759
i35 -53.034873
EXT_SOURCE_3 -26.989443 
shap_new.tail()
SHAP
i11 51.871609
i10 106.712072
i9 184.370958
i8 214.506221
i7 229.400134 
.to_frame().T.head()
DAYS_BIRTH NAME_CONTRACT_STATUS_Refused_mean DAYS_CREDIT_ENDDATE_mean AMT_INSTALMENT_max DAYS_DECISION_mean AMT_CREDIT_mean ORGANIZATION_TYPE_MEAN AMT_DOWN_PAYMENT_max AMT_ANNUITY AMT_CREDIT_MAX_OVERDUE_max ... CODE_REJECT_REASON_HC_mean CNT_PAYMENT_max CNT_INSTALMENT_FUTURE_mean EXT_SOURCE_1 DAYS_CREDIT_ENDDATE_max EXT_SOURCE_3 CNT_INSTALMENT_FUTURE_max STATUS_1_mean CREDIT_ACTIVE_Closed_mean NUM_INSTALMENT_NUMBER_max
0 -94.329393 -73.081732 -59.153025 -59.117657 -57.903585 -48.936621 -42.214968 -40.645964 -39.254812 -30.469348 ... 30.599336 31.054876 32.973514 33.522551 37.238967 40.022553 40.582652 50.796038 61.407426 76.074902

1 rows × 616 columns

shap.summary_plot(shap_values, X_fit)

png

DAYS_BIRTH , DAYS_ID_PUBLISH

feat_imp = feature_importances.groupby("feature").mean().sort_values("importance", ascending=False)

feature_importances.groupby("feature").mean().sort_values("importance", ascending=False)
importance fold
feature
i8 81.6 3.0
i12 72.2 3.0
i11 71.0 3.0
i17 61.0 3.0
i14 59.6 3.0
i10 58.6 3.0
i24 58.6 3.0
i3 58.2 3.0
i127 55.2 3.0
i9 55.2 3.0
i51 54.8 3.0
i15 52.6 3.0
i136 52.0 3.0
inst_pay 52.0 3.0
i23 51.2 3.0
DAYS_BIRTH 51.0 3.0
i175 49.2 3.0
i18 48.0 3.0
i43 45.4 3.0
i13 45.0 3.0
i28 44.6 3.0
i35 41.6 3.0
i2 41.4 3.0
i20 40.4 3.0
i119 40.0 3.0
i37 39.4 3.0
i6 38.8 3.0
DAYS_CREDIT_max 38.4 3.0
i1 37.8 3.0
i404 37.2 3.0
... ... ...
CNT_PAYMENT_max 1.6 3.0
CNT_CHILDREN 1.4 3.0
NAME_HOUSING_TYPE_MEAN 1.4 3.0
NAME_CASH_LOAN_PURPOSE_Buying a used car_mean 1.4 3.0
AMT_DRAWINGS_ATM_CURRENT_max 1.4 3.0
CHANNEL_TYPE_Contact center_mean 1.4 3.0
PRODUCT_COMBINATION_Cash Street: middle_mean 1.4 3.0
ENTRANCES_AVG 1.2 3.0
AMT_DRAWINGS_ATM_CURRENT_mean 1.2 3.0
AMT_INST_MIN_REGULARITY_mean 1.0 3.0
i417 1.0 3.0
PRODUCT_COMBINATION_Card X-Sell_mean 1.0 3.0
BASEMENTAREA_MODE 1.0 3.0
FLOORSMIN_AVG 1.0 3.0
i79 1.0 3.0
MONTHS_BALANCE_mean 0.8 3.0
LIVINGAPARTMENTS_AVG 0.8 3.0
i181 0.8 3.0
i116 0.8 3.0
DAYS_TERMINATION_max 0.8 3.0
NFLAG_INSURED_ON_APPROVAL_mean 0.8 3.0
i495 0.6 3.0
NAME_CASH_LOAN_PURPOSE_Wedding / gift / holiday_max 0.6 3.0
i428 0.6 3.0
CNT_FAM_MEMBERS 0.6 3.0
i455 0.4 3.0
i212 0.4 3.0
AMT_DRAWINGS_CURRENT_mean 0.2 3.0
i310 0.0 3.0
NAME_SELLER_INDUSTRY_Jewelry_mean 0.0 3.0

433 rows × 2 columns