fl4health.feature_alignment.feature_type_extraction module¶

class FeatureMeta(**kwargs)[source]¶

Get the feature type.

update(meta)[source]¶

Update meta attributes.

Parameters:: meta (list[tuple[str, Any]]) – List of tuples in the format (attribute name, attribute value).
Return type:: None

class Features(data, features, by=None, targets=None, force_types=None)[source]¶

__init__(data, features, by=None, targets=None, force_types=None)[source]¶

Features.

Parameters:

data (pd.DataFrame) – Features data.
features (str | list[str]) – List of feature columns. The remaining columns are treated as metadata.
by (str | list[str] | None, optional) – Columns to groupby during processing, affecting how the features are treated. Defaults to None.
targets (str | list[str] | None, optional) – Column names to specify as target features. Defaults to None.
force_types (dict[str, FeatureType] | None, optional) – Mapping of column names to type. These columns are forced to be of the specified type. Defaults to None.

property types: dict[str, FeatureType]¶

Access as attribute, feature type names.

NOTE: These are framework-specific feature names.

class TabularFeatures(data, features, by, targets=None, force_types=None)[source]¶

__init__(data, features, by, targets=None, force_types=None)[source]¶

Tabular features.

Parameters:

data (pd.DataFrame) – Data for the table
features (str | list[str]) – List of feature columns. The remaining columns are treated as metadata.
by (str) – Columns to groupby during processing, affecting how the features are treated.
targets (str | list[str] | None, optional) – Column names to specify as target features. Defaults to None.
force_types (dict[str, FeatureType] | None, optional) – Mapping of column names to type. These columns are forced to be of the specified type. Defaults to None.

Raises:

ValueError – Tabular features index input as a string representing a column

has_columns(data, cols, exactly=False, raise_error=False)[source]¶

Check if data has required columns for processing.

Parameters:

data (pd.DataFrame) – DataFrame to check.
cols (str | list[str]) – List of column names that must be present in data.
exactly (bool, optional) – Whether columns need to be an exact match. Defaults to False.
raise_error (bool, optional) – Whether to raise a ValueError if there are missing columns. Defaults to False.

Raises:

ValueError – Missing required columns.
ValueError – Must have exactly the columns, will throw if not and exactly is True.

Returns:

True if all required columns are present, otherwise False.

Return type:

bool

to_list(obj)[source]¶

Convert some object to a list of object(s) unless already one.