FeatureSpec#

class scikitplot.feature_selection.FeatureSpec(name, position, role, family, kind, kind_origin, stats, assessments=(), tags=(), binning=None, aliases=())[source]#

Everything a fitted schema knows about one column.

Parameters:

namestr: Current column name.
positionint: Column position in the training frame.
rolestr: FeatureRole value.
familystr: DtypeFamily value.
kindstr: FeatureKind value.
kind_origin{“inferred”, “declared”}: Where the kind came from.
statsFeatureStats: Training statistics.
assessmentstuple of Assessment: Sorted by name.
tagstuple of str, default=(): User tags.
binningBinningRecommendation or None, default=None: Bin recommendation, when binning is enabled.
aliasestuple of str, default=(): Earlier names, oldest first (see FeatureSchema.rename).

Parameters:

name (str)
position (int)
role (str)
family (str)
kind (str)
kind_origin (str)
stats (FeatureStats)
assessments (tuple[Assessment, ...])
tags (tuple[str, ...])
binning (BinningRecommendation | None)
aliases (tuple[str, ...])

See also

FeatureSchema: The collection of specs.

aliases: tuple[str, ...] = ()#

assess(name)[source]#

Return the recorded answer to assessment name.

Parameters:

namestr: Assessment name ("skewed", "discrete" …).

Returns:

Assessment: State, reasons, evidence and provenance.

Raises:

KeyError: If the schema’s policy did not run that assessment; the message lists the available names.

Parameters:

name (str)

Return type:

property assessment_names: tuple[str, ...]#

Names of the recorded assessments.

Type:: Tuple of str

assessments: tuple[Assessment, ...] = ()#

binning: BinningRecommendation | None = None#

explain()[source]#

Return a short, human-readable description of the feature.

Returns:

str: Several lines: role, kind, key statistics, true assessments, and the binning recommendation.

Return type:

family: str[source]#

classmethod from_dict(data)[source]#

Rebuild a spec from to_dict output, checking every field.

Parameters:

datamapping: Stored form.

Returns:

FeatureSpec: The spec.

Raises:

ValueError: For an off-contract record.

Parameters:

data (Mapping[str, Any])

Return type:

is_(name)[source]#

Answer “am I name?” strictly.

Parameters:

namestr: Assessment name.

Returns:

bool: True only when the state is "true"; unknown and not_applicable give False. Use state to tell them apart.

Parameters:

name (str)

Return type:

property is_model_input: bool#

The role is "feature".

Type:: bool

kind: str[source]#

kind_origin: str[source]#

name: str[source]#

position: int[source]#

role: str[source]#

state(name)[source]#

Return the state of assessment name.

Parameters:

namestr: Assessment name.

Returns:

str: "true", "false", "unknown" or "not_applicable".

Parameters:

name (str)

Return type:

stats: FeatureStats[source]#

tags: tuple[str, ...] = ()#

to_dict()[source]#

Return the shared-data-model form.

Return type:: dict[str, Any]