Currently, the LofoImportance class doesn't support cross-validation with groups. For example, if we work with sales demand forecasting and data contains multiple items, then each validation data should contain sold items, i,e. groups. Therefore, the purpose of this PR is to enable LofoImportance to work directly with sklearn's GroupKFold or mlextend's GroupTimeSeriesSplit as cross_validate function from sklearn supports passing groups argument.
Currently, the LofoImportance class doesn't support cross-validation with groups. For example, if we work with sales demand forecasting and data contains multiple items, then each validation data should contain sold items, i,e. groups. Therefore, the purpose of this PR is to enable LofoImportance to work directly with sklearn's GroupKFold or mlextend's GroupTimeSeriesSplit as cross_validate function from sklearn supports passing groups argument.