Association#

synthesized.common.rules.Association

class Association(binding_mask, associations=None, nan_associations=None)#

Define a relationship between two categorical variables that the synthesizer must obey.

The index to value mapping is determined from the order of the categories in the appropriate meta class

Parameters
  • binding_mask (np.ndarray) – boolean mask representing the allowed associations between categorical values.

  • associations (List[str]) – list of column names to associate.

  • nan_associations (List[str]) – list of nan indicator columns to associate.

Methods

detect_association(df, df_meta[, ...])

Constructor that automatically generates a binding mask for the association based on a dataframe and masks all combinations of inputs that don't appear in the dataframe.

classmethod detect_association(df, df_meta, associations=None, nan_associations=None)#

Constructor that automatically generates a binding mask for the association based on a dataframe and masks all combinations of inputs that don’t appear in the dataframe.

Parameters
  • df (pd.DataFrame) – input DataFrame.

  • df_meta (DataFrameMeta) – extracted DataFrameMeta

  • associations (List[str]) – list of regular columns to check for associations.

  • nan_associations (List[str]) – list of columns to check nan value associations.

Returns

new Association object with automatically generated binding mask.

Return type

association