Preprocessors

TabularPreproc

from_schema classmethod

from_schema(
    schema: Schema,
    ctx_cols: dict[str, Sequence[str]] | None = None,
    preprocessors: dict[
        str, dict[str, ColumnPreproc | ArColumn | None]
    ]
    | None = None,
) -> TabularPreproc

Build a preprocessor for tabular data from the Schema.

Parameters:

Name	Type	Description	Default
schema	Schema	A Schema object.	required
ctx_cols	dict[str, Sequence[str]] | None	A dictionary with the columns to be used as context. May contain: 1. Only the root table as key and a subset of its columns as value. 2. All the tables as keys and a subset of each table's columns as values.	None
preprocessors	dict[str, dict[str, ColumnPreproc | ArColumn | None]] | None	A dictionary containing preprocessing instructions for each column in the schema. Keys are table names, values are dictionaries with column names as keys and preprocessing instructions as values. Preprocessing instructions can be instances of ColumnPreproc, a column preprocessor, or None. If None, the column will be ignored. For the columns for which a preprocessor is not provided, the default preprocessor will be instantiated based on the Column type defined in the Schema.	None

Returns:

Type	Description
TabularPreproc	A TabularPreproc object.

fit

fit(data: RelationalData) -> TabularPreproc

Fit the preprocessor to the given RelationalData.

Parameters:

Name	Type	Description	Default
data	RelationalData	The RelationalData to fit the preprocessor to.	required

Returns:

Type	Description
TabularPreproc	The fitted TabularPreproc object.

select_ctx

select_ctx(
    data: RelationalData, idx: Sequence[int] | None = None
) -> RelationalData

Select the context from the input data.

Parameters:

Name	Type	Description	Default
data	RelationalData	The RelationalData from which to extract the context.	required
idx	Sequence[int] | None	The indices of the root table to select. May be repeated. If None, all indices will be taken once and the keys will be kept. Otherwise, the keys will be reset.	None

Returns:

Type	Description
RelationalData	A RelationalData with the selected context.

sample_ctx

sample_ctx(
    data: RelationalData,
    n_samples: int | None = None,
    rng: Generator | int | None = None,
) -> RelationalData

Sample the context from th einput data.

Parameters:

Name	Type	Description	Default
data	RelationalData	The RelationalData from which to extract the context.	required
n_samples	int | None	The number of context samples. If None, the number of samples will be equal to the number of samples in the root table.	None
rng	Generator | int | None	A np.random.Generator or an integer seed to control the randomness during sampling. If None, a random seed is generated.	None

Returns:

Type	Description
RelationalData	A RelationalData with the sampled context.

TextPreproc

from_schema_table classmethod

from_schema_table(
    schema: Schema, table: str
) -> TextPreproc

Build a preprocessor for the text columns of a table from the Schema.

Parameters:

Name	Type	Description	Default
schema	Schema	A Schema object.	required
table	str	Name of the target table in the schema that contains text columns.	required

Returns:

Type	Description
TextPreproc	A TextPreproc object.

from_tabular classmethod

from_tabular(
    preproc: TabularPreproc[_AP], table: str
) -> TextPreproc

Build a preprocessor for the text columns of a table from the TabularPreproc used for the tabular data.

Parameters:

Name	Type	Description	Default
preproc	TabularPreproc[_AP]	A TabularPreproc object used for the tabular part of the data.	required
table	str	Name of the target table in the schema that contains text columns.	required

Returns:

Type	Description
TextPreproc	A TextPreproc object.

fit

fit(data: RelationalData) -> TextPreproc

Fit the preprocessor to the given RelationalData.

Parameters:

Name	Type	Description	Default
data	RelationalData	The RelationalData to fit the preprocessor to.	required

Returns:

Type	Description
TextPreproc	The fitted TextPreproc object.

ColumnPreproc dataclass

Preprocessing instructions for a column.

Parameters:

Name	Type	Description	Default
special_values	Sequence | None	A sequence of special values to handle during preprocessing.	None
impute_nan	bool | None	A flag indicating whether to impute NaN values during preprocessing. If True, NaN values will not be sampled during the generation of synthetic data.	None
non_sample_values	Sequence | None	A sequence of values that should not be sampled during the generation of synthetic data.	None
protection	Protection | bool | None	A Protection object or boolean flag indicating whether to apply protection to the column. If boolean, the default protection is applied, otherwise the Protection object configures the protection.	None