check_continuous.py

import datetime

cf_history = [('CF_B_HST_MONTH_24', 'CF_B_HST_OS_AMOUNT_24', 'CF_B_HST_AC_DPD_24'),
 ('CF_B_HST_MONTH_23', 'CF_B_HST_OS_AMOUNT_23', 'CF_B_HST_AC_DPD_23'),
 ('CF_B_HST_MONTH_22', 'CF_B_HST_OS_AMOUNT_22', 'CF_B_HST_AC_DPD_22'),
 ('CF_B_HST_MONTH_21', 'CF_B_HST_OS_AMOUNT_21', 'CF_B_HST_AC_DPD_21'),
 ('CF_B_HST_MONTH_20', 'CF_B_HST_OS_AMOUNT_20', 'CF_B_HST_AC_DPD_20'),
 ('CF_B_HST_MONTH_19', 'CF_B_HST_OS_AMOUNT_19', 'CF_B_HST_AC_DPD_19'),
 ('CF_B_HST_MONTH_18', 'CF_B_HST_OS_AMOUNT_18', 'CF_B_HST_AC_DPD_18'),
 ('CF_B_HST_MONTH_17', 'CF_B_HST_OS_AMOUNT_17', 'CF_B_HST_AC_DPD_17'),
 ('CF_B_HST_MONTH_16', 'CF_B_HST_OS_AMOUNT_16', 'CF_B_HST_AC_DPD_16'),
 ('CF_B_HST_MONTH_15', 'CF_B_HST_OS_AMOUNT_15', 'CF_B_HST_AC_DPD_15'),
 ('CF_B_HST_MONTH_14', 'CF_B_HST_OS_AMOUNT_14', 'CF_B_HST_AC_DPD_14'),
 ('CF_B_HST_MONTH_13', 'CF_B_HST_OS_AMOUNT_13', 'CF_B_HST_AC_DPD_13'),
 ('CF_B_HST_MONTH_12', 'CF_B_HST_OS_AMOUNT_12', 'CF_B_HST_AC_DPD_12'),
 ('CF_B_HST_MONTH_11', 'CF_B_HST_OS_AMOUNT_11', 'CF_B_HST_AC_DPD_11'),
 ('CF_B_HST_MONTH_10', 'CF_B_HST_OS_AMOUNT_10', 'CF_B_HST_AC_DPD_10'),
 ('CF_B_HST_MONTH_9', 'CF_B_HST_OS_AMOUNT_9', 'CF_B_HST_AC_DPD_9'),
 ('CF_B_HST_MONTH_8', 'CF_B_HST_OS_AMOUNT_8', 'CF_B_HST_AC_DPD_8'),
 ('CF_B_HST_MONTH_7', 'CF_B_HST_OS_AMOUNT_7', 'CF_B_HST_AC_DPD_7'),
 ('CF_B_HST_MONTH_6', 'CF_B_HST_OS_AMOUNT_6', 'CF_B_HST_AC_DPD_6'),
 ('CF_B_HST_MONTH_5', 'CF_B_HST_OS_AMOUNT_5', 'CF_B_HST_AC_DPD_5'),
 ('CF_B_HST_MONTH_4', 'CF_B_HST_OS_AMOUNT_4', 'CF_B_HST_AC_DPD_4'),
 ('CF_B_HST_MONTH_3', 'CF_B_HST_OS_AMOUNT_3', 'CF_B_HST_AC_DPD_3'),
 ('CF_B_HST_MONTH_2', 'CF_B_HST_OS_AMOUNT_2', 'CF_B_HST_AC_DPD_2'),
 ('CF_B_HST_MONTH_1', 'CF_B_HST_OS_AMOUNT_1', 'CF_B_HST_AC_DPD_1')]
dpd_delinquency_1 = ['SMA-2', 'NPA',]
dpd_delinquency_2 = ['SMA-2', 'NPA', 'Other']


def check_if_valid_sma(row, dpd_delinquency, n_months=12):
    def check_if_continuous_date_available(index):
        delinquency_flag_inside = False
        continuous_flag_inside = False
        delinquency_flag_outside = False
        
        next_n_months = index + n_months + 1
        i = 0
        for date_col, amount_col, dpd_col in cf_history[0: next_n_months]:
            date = row[date_col]
            amount = row[amount_col]
            dpd = row[dpd_col]
            
            if i < index:
                if dpd in dpd_delinquency:
                    delinquency_flag_outside = True
                i = i + 1
                continue
            
            if dpd in dpd_delinquency:
#                 print(dpd)
                delinquency_flag_inside = True
            if date == np.nan or dpd == '-' or dpd == '' or amount == np.nan or date < datetime.datetime(year=2015, month=1, day=1):
                continuous_flag_inside = False
                return False, delinquency_flag_inside, delinquency_flag_outside 
            i = i + 1
                
        return True, delinquency_flag_inside, delinquency_flag_outside
            
            
    break_flag = False
    continuous_flag = False
    delinquency_flag = False
    delinquency_flag_outside = False
    for i, (date_col, amount_col, dpd_col) in enumerate(cf_history):
        n_month = i + 1
        is_continuous, is_delinquent_inside, is_delinquent_outside = check_if_continuous_date_available(i)
        if is_continuous:
            continuous_flag = True
            delinquency_flag = is_delinquent_inside
            delinquency_flag_outside = is_delinquent_outside
            break
        if n_month > len(cf_history) - n_months:
            break
#             pass
            
    return continuous_flag, delinquency_flag, is_delinquent_outside


df_cf['is_valid_1'] = df_cf.apply(lambda x: check_if_valid_sma(x, dpd_delinquency=dpd_delinquency_1), axis=1)
df_cf['is_valid_2'] = df_cf.apply(lambda x: check_if_valid_sma(x, dpd_delinquency=dpd_delinquency_2), axis=1)