param = {
'colsample_bytree': 0.8,
'subsample': 0.1,
'learning_rate': 0.1,
'metric': 'auc',
'max_depth': 4,
'objective': 'binary',
'nthread': 64,
'seed': SEED
}
gc.collect()
yhat, imp, ret = ex.stacking(X,
y,
param,
NROUND,
nfold=5,
esr=50,
categorical_feature=categorical_feature)
t = datetime.today()
date = t.date()
hour = t.hour
imp.to_csv('imp_{}-{:02d}h.csv'.format(date, hour), index=False)
# =============================================================================
# cv
# =============================================================================
#model = xgb.train(param, dbuild, NROUND, watchlist, verbose_eval=10,
# early_stopping_rounds=50)
'max_bin': 100,
'colsample_bytree': 0.5,
'subsample': 0.5,
'nthread': 64,
'bagging_freq': 1,
'seed': SEED
}
categorical_feature = [
'NAME_CONTRACT_TYPE', 'CODE_GENDER', 'FLAG_OWN_CAR', 'FLAG_OWN_REALTY',
'NAME_TYPE_SUITE', 'NAME_INCOME_TYPE', 'NAME_EDUCATION_TYPE',
'NAME_FAMILY_STATUS', 'NAME_HOUSING_TYPE', 'OCCUPATION_TYPE',
'WEEKDAY_APPR_PROCESS_START', 'ORGANIZATION_TYPE', 'FONDKAPREMONT_MODE',
'HOUSETYPE_MODE', 'WALLSMATERIAL_MODE', 'EMERGENCYSTATE_MODE'
]
dtrain = lgb.Dataset(X, y, categorical_feature=categorical_feature)
yhat, imp, ret = ex.stacking(X,
y,
param,
9999,
esr=50,
seed=SEED,
categorical_feature=categorical_feature)
imp.to_csv(f'LOG/imp_{__file__}.csv', index=False)
#==============================================================================
utils.end(__file__)