featminer — Predictive Power Report

Feature generation time: 1m 15.6s

26 features
AUC: strong > 0.6 medium 0.55–0.6 weak < 0.55
PSI: stable < 0.1 medium 0.1–0.2 unstable ≥ 0.2
Nulls: full coverage 0% partial coverage 0–50% low coverage > 50%
Dataset summary
input_rows1465325
input_columns17
rows_after_sampling1465325
columns_after_preprocessing17
targetTARGET
aggregation_keySK_ID_CURR
target_entities263491
baseline_features0
generated_features26
Algorithm parameters
max_filter_conditions2
min_category_fraction0.01
thr_auc0.55
thr_corr0.7
thr_nulls0.5
thr_nonunique0.95
max_cells_threshold100000000
formula_setstable
formulasFraction, Count, Last, Max, Sum, Mean, Min, Nunique, Std
⚙
Feature name AUC Nulls % PSI Python code
days_credit__mean0.60300.0%0.0001df['DAYS_CREDIT'].groupby(df['SK_ID_CURR']).mean()
days_credit__credit_type__credit_card__mean_e03b83de0.592834.7%0.0002df['DAYS_CREDIT'][(df['CREDIT_TYPE'] == 'Credit card')].groupby(df['SK_ID_CURR']).mean()
days_credit__credit_type__consumer_credit__mean_9e4167de0.58926.3%0.0002df['DAYS_CREDIT'][(df['CREDIT_TYPE'] == 'Consumer credit')].groupby(df['SK_ID_CURR']).mean()
credit_active__closed__fraction0.58610.0%0.0001((df['CREDIT_ACTIVE'] == 'Closed')).groupby(df['SK_ID_CURR']).mean()
amt_credit_sum_debt__credit_type__credit_card__sum_00e9a1470.583234.7%0.0001df['AMT_CREDIT_SUM_DEBT'][(df['CREDIT_TYPE'] == 'Credit card')].groupby(df['SK_ID_CURR']).sum()
days_credit__max0.58030.0%0.0002df['DAYS_CREDIT'].groupby(df['SK_ID_CURR']).max()
days_credit_update__credit_type__credit_card__max_2d1a15b20.577734.7%0.0002df['DAYS_CREDIT_UPDATE'][(df['CREDIT_TYPE'] == 'Credit card')].groupby(df['SK_ID_CURR']).max()
days_credit_enddate__sum0.57390.0%0.0001df['DAYS_CREDIT_ENDDATE'].groupby(df['SK_ID_CURR']).sum()
days_credit__credit_active__active__mean0.572717.6%0.0002df['DAYS_CREDIT'][(df['CREDIT_ACTIVE'] == 'Active')].groupby(df['SK_ID_CURR']).mean()
days_credit_update__min0.56920.0%0.0000df['DAYS_CREDIT_UPDATE'].groupby(df['SK_ID_CURR']).min()
days_enddate_fact__sum0.56810.0%0.0001df['DAYS_ENDDATE_FACT'].groupby(df['SK_ID_CURR']).sum()
days_credit_update__credit_type__consumer_credit__min_d115248d0.56776.3%0.0002df['DAYS_CREDIT_UPDATE'][(df['CREDIT_TYPE'] == 'Consumer credit')].groupby(df['SK_ID_CURR']).min()
days_credit_enddate__min0.56690.9%0.0001df['DAYS_CREDIT_ENDDATE'].groupby(df['SK_ID_CURR']).min()
days_credit_enddate__credit_type__consumer_credit__max_99cfc7c90.56116.6%0.0002df['DAYS_CREDIT_ENDDATE'][(df['CREDIT_TYPE'] == 'Consumer credit')].groupby(df['SK_ID_CURR']).max()
credit_active__active__count0.56040.0%0.0001((df['CREDIT_ACTIVE'] == 'Active')).groupby(df['SK_ID_CURR']).sum()
days_credit__std0.557813.7%0.0001df['DAYS_CREDIT'].groupby(df['SK_ID_CURR']).std()
amt_credit_sum_debt__mean0.55752.8%0.0001df['AMT_CREDIT_SUM_DEBT'].groupby(df['SK_ID_CURR']).mean()
amt_credit_sum_debt__credit_type__credit_card__min_328ec1af0.556138.9%0.0001df['AMT_CREDIT_SUM_DEBT'][(df['CREDIT_TYPE'] == 'Credit card')].groupby(df['SK_ID_CURR']).min()
amt_credit_sum_debt__credit_type__consumer_credit__mean_24a83e390.554910.1%0.0002df['AMT_CREDIT_SUM_DEBT'][(df['CREDIT_TYPE'] == 'Consumer credit')].groupby(df['SK_ID_CURR']).mean()
credit_active__active__credit_type__consumer_credit__fraction_3b00388f0.55490.0%0.0002((df['CREDIT_ACTIVE'] == 'Active') & (df['CREDIT_TYPE'] == 'Consumer credit')).groupby(df['SK_ID_CURR']).mean()
days_credit_update__credit_active__closed__min0.554812.6%0.0001df['DAYS_CREDIT_UPDATE'][(df['CREDIT_ACTIVE'] == 'Closed')].groupby(df['SK_ID_CURR']).min()
amt_credit_sum__credit_active__active__min0.552617.6%0.0001df['AMT_CREDIT_SUM'][(df['CREDIT_ACTIVE'] == 'Active')].groupby(df['SK_ID_CURR']).min()
amt_credit_sum_debt__sum0.55140.0%0.0001df['AMT_CREDIT_SUM_DEBT'].groupby(df['SK_ID_CURR']).sum()
days_credit_enddate__credit_type__credit_card__min_9827bf250.551242.7%0.0001df['DAYS_CREDIT_ENDDATE'][(df['CREDIT_TYPE'] == 'Credit card')].groupby(df['SK_ID_CURR']).min()
credit_active__active__credit_type__credit_card__fraction_6037a9ac0.55050.0%0.0001((df['CREDIT_ACTIVE'] == 'Active') & (df['CREDIT_TYPE'] == 'Credit card')).groupby(df['SK_ID_CURR']).mean()
days_credit_update__credit_type__consumer_credit__max_e0df8a570.55046.3%0.0002df['DAYS_CREDIT_UPDATE'][(df['CREDIT_TYPE'] == 'Consumer credit')].groupby(df['SK_ID_CURR']).max()