At the moment DataFrameWithInfo takes as parameters metadata_cols (tuple, i.e. the columns of the dataframe to be used as metadata) and metadata_as_features (bool, i.e. whether to consider all the metadata columns as features). It could be useful to set only a subset of the metadata columns as features, or even to exclude some columns both from the metadata and the features sets.
To accomplish these goals, DataFrameWithInfo could accept both metadata_cols and feature_cols parameters as tuples of strings enumerating the metadata and the feature columns, respectively. I suggest feature_cols to be an optional parameter, by default equal to None, meaning that all the columns but the metadata columns are features columns.
At the moment
DataFrameWithInfotakes as parametersmetadata_cols(tuple, i.e. the columns of the dataframe to be used as metadata) andmetadata_as_features(bool, i.e. whether to consider all the metadata columns as features). It could be useful to set only a subset of the metadata columns as features, or even to exclude some columns both from the metadata and the features sets.To accomplish these goals,
DataFrameWithInfocould accept bothmetadata_colsandfeature_colsparameters as tuples of strings enumerating the metadata and the feature columns, respectively. I suggestfeature_colsto be an optional parameter, by default equal toNone, meaning that all the columns but the metadata columns are features columns.