[Python] exogene and static features for lightgbm in automlforecast

i tried several approaches to implement exogene features in my lightgbm prediction with automlforecast and optuna. I also read the documentation, but i do not find the right positon where to implement static_features = []

As there are no static_features, all except y, ds and unique_id are handled as dynamic/exogene features.

Actually with all implementations I get "TypeError: AutoMLForecast.init() got an unexpected keyword argument 'static_features'"

That is part of my code:

def my_lgb_config(trial: optuna.Trial):
return {
'learning_rate': trial.suggest_float('learning_rate', 0.01, 0.1),
'verbosity': -1, # keine Infos
'num_leaves': trial.suggest_int('num_leaves', 2, 2048, log=True),
'objective': trial.suggest_categorical('objective', ['regression_l1', 'mape','regression', 'huber','fair']),
'boosting_type': trial.suggest_categorical('boosting_type', ['gbdt', 'dart', 'goss', 'rf']),
'max_depth': trial.suggest_int('max_depth', 3, 50), #org 3, 12
'min_child_weight': trial.suggest_float('min_child_weight', 0.001, 10.0),
'min_child_samples': trial.suggest_int('min_child_samples', 5, 100),
'subsample': trial.suggest_float('subsample', 0.5, 1.0),
'colsample_bytree': trial.suggest_float('colsample_bytree', 0.5, 1.0),
'reg_alpha': trial.suggest_float('reg_alpha', 0.0, 1.0),
'reg_lambda': trial.suggest_float('reg_lambda', 0.0, 1.0),
'scale_pos_weight': trial.suggest_float('scale_pos_weight', 0.1, 10.0),
'n_estimators': trial.suggest_int('n_estimators', 50, 1000), #50, 1000
'max_bin': trial.suggest_int('max_bin', 100, 500),
'metric': trial.suggest_categorical('metric',['rmse','mae','mape']),
#'static_features': trial.suggest_categorical('static_features', [])
}

my_lgb = AutoModel(
model=lgb.LGBMRegressor(random_state=0),
config=my_lgb_config)

fcst = AutoMLForecast(
models={'my_lgb': my_lgb},
freq='D',
season_length=7,
static_features=[],
)


Hope you can help me to find the right position.

thanks in advanced!

Continue reading...