Documentation of decision_rules

decision_rules.conditions

Contains logical conditions classes.

class decision_rules.conditions.AttributesCondition(column_left: int, column_right: int, operator: str)

Bases: AbstractCondition

Condition specifying a relation between two attributes (columns). Possible relations are: =, !=, >, <

Example in rule:: IF petal_length > sepal_width THEN y = 1

property attributes: frozenset[int]: Returns: frozenset[int]: condition attributes

to_string(columns_names: str) → str

Parameters:: columns_names (list[str])
Returns:: condition string representation
Return type:: str

class decision_rules.conditions.CompoundCondition(subconditions: list[ElementaryCondition], logic_operator: LogicOperators = LogicOperators.CONJUNCTION)

Bases: AbstractCondition

Condition specyfing logical conjunction or alternative of other conditions.

Example in rule:: IF petal_length = <-0.015, 0.185> AND sepal_width = <-0.047, 0.253> THEN y = 1

property attributes: frozenset[int]: Returns: frozenset[int]: condition attributes

to_string(columns_names: list[str]) → str

Parameters:: columns_names (list[str])
Returns:: condition string representation
Return type:: str

class decision_rules.conditions.ElementaryCondition(column_index: int, left: float = -inf, right: float = inf, left_closed: bool = False, right_closed: bool = False)

Bases: AbstractCondition

Numerical attributes condition.

It has left and right boundary and takes following form:: {LEFT} < {ATTRIBUTE} < {RIGHT}
Example in rule:: IF petal_length = <-0.015, 0.185> THEN y = 1

property attributes: frozenset[int]: Returns: frozenset[int]: condition attributes

to_string(columns_names: str) → str

Parameters:: columns_names (list[str])
Returns:: condition string representation
Return type:: str

class decision_rules.conditions.LogicOperators(value, names=None, *values, module=None, qualname=None, type=None, start=1, boundary=None)

Bases: Enum

ALTERNATIVE = 'ALTERNATIVE'

CONJUNCTION = 'CONJUNCTION'

class decision_rules.conditions.NominalCondition(column_index: int, value: str)

Bases: AbstractCondition

Class for elementary condition for nominal attributes.

It takes following form:

IF {ATTRIBUTE} = {VALUE} THEN y = {DECISION}

Example

IF gender = 1 THEN y = 1

property attributes: frozenset[int]: Returns: frozenset[int]: condition attributes

to_string(columns_names: list[str]) → str

Parameters:: columns_names (list[str])
Returns:: condition string representation
Return type:: str

decision_rules.histogram

class decision_rules.histogram.Histograms(*, min: int, max: int, bin_edges: list[float], histograms: dict[str, list[int]])

Bases: BaseModel

bin_edges: list[float]

histograms: dict[str, list[int]]

max: int

min: int

model_config: ClassVar[ConfigDict] = {}: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

model_fields: ClassVar[dict[str, FieldInfo]] = {'bin_edges': FieldInfo(annotation=list[float], required=True), 'histograms': FieldInfo(annotation=dict[str, list[int]], required=True), 'max': FieldInfo(annotation=int, required=True), 'min': FieldInfo(annotation=int, required=True)}

Metadata about the fields defined on the model, mapping of field names to [FieldInfo][pydantic.fields.FieldInfo].

This replaces Model.__fields__ from Pydantic V1.

decision_rules.histogram.get_histograms(model: AbstractRuleSet, dataset: DataFrame, problem_type: ProblemTypes, bins: int, for_rules: list[str] = ()) → Histograms

decision_rules.measures

Contains functions for calculating different qualities measures. They can be used for calculating rule qualities and voting weights

decision_rules.measures.accuracy(c: Coverage) → float

decision_rules.measures.c1(c: Coverage) → float

decision_rules.measures.c2(c: Coverage) → float

decision_rules.measures.c_foil(c: Coverage) → float

decision_rules.measures.cn2_significnce(c: Coverage) → float

decision_rules.measures.confidence(c: Coverage) → float

decision_rules.measures.correlation(c: Coverage) → float

decision_rules.measures.coverage(c: Coverage) → float

decision_rules.measures.f_bayesian_confirmation(c: Coverage) → float

decision_rules.measures.f_measure(c: Coverage) → float

decision_rules.measures.full_coverage(c: Coverage) → float

decision_rules.measures.g_measure(c: Coverage) → float

decision_rules.measures.geo_rss(c: Coverage) → float

decision_rules.measures.information_gain(c: Coverage) → float

decision_rules.measures.j_measure(c: Coverage) → float

decision_rules.measures.kappa(c: Coverage) → float

decision_rules.measures.klosgen(c: Coverage) → float

decision_rules.measures.laplace(c: Coverage) → float

decision_rules.measures.lift(c: Coverage) → float

decision_rules.measures.logical_sufficiency(c: Coverage) → float

decision_rules.measures.m_estimate(c: Coverage) → float

decision_rules.measures.mutual_support(c: Coverage) → float

decision_rules.measures.novelty(c: Coverage) → float

decision_rules.measures.odds_ratio(c: Coverage) → float

decision_rules.measures.one_way_support(c: Coverage) → float

decision_rules.measures.pawlak_dependency_factor(c: Coverage) → float

decision_rules.measures.precision(c: Coverage) → float

decision_rules.measures.q2(c: Coverage) → float

decision_rules.measures.relative_risk(c: Coverage) → float

decision_rules.measures.ripper(c: Coverage) → float

decision_rules.measures.rss(c: Coverage) → float

decision_rules.measures.rule_interest(c: Coverage) → float

decision_rules.measures.s_bayesian(c: Coverage) → float

decision_rules.measures.sensitivity(c: Coverage) → float

decision_rules.measures.specificity(c: Coverage) → float

decision_rules.measures.two_way_support(c: Coverage) → float

decision_rules.measures.weighted_laplace(c: Coverage) → float

decision_rules.measures.weighted_relative_accuracy(c: Coverage) → float

decision_rules.measures.yails(c: Coverage) → float

decision_rules.problem

class decision_rules.problem.ProblemTypes(value, names=None, *values, module=None, qualname=None, type=None, start=1, boundary=None)

Bases: str, Enum

CLASSIFICATION = 'classification'

REGRESSION = 'regression'

SURVIVAL = 'survival'

decision_rules.settings

Contains global configuration conditions classes.

decision_rules.settings.FLOAT_DISPLAY_PRECISION

controls floating points numbers precision when displaying rules as strings.

Type:: str

decision_rules.settings.CONCISE_NUMERICAL_CONDITIONS_FORM

If set to true, string representations of numerical elementary conditions where one of the interval boundary is +/- infinity will be writen in more concise form of “X < right” or “X > left” instead of the full form: “X = <left, right)”

Type:: bool