biogeme.latent_variables.model_spec module

class biogeme.latent_variables.model_spec.IndicatorMeasurementSpec(indicator_name, measurement_model, measurement_sigma=None)

Bases: object

Measurement specification attached to one indicator.

This object describes how one semantic indicator is modeled in a given run, independently of the indicator definition itself.

Parameters:

• indicator_name (str) – Name of the indicator being modeled.

• measurement_model (MeasurementModel) – Statistical measurement model used for that indicator.

• measurement_sigma (PositiveParameterSpec | None) – Optional specification of the measurement scale on its natural scale.

indicator_name: str

measurement_model: MeasurementModel

measurement_sigma: PositiveParameterSpec | None

class biogeme.latent_variables.model_spec.LatentVariable(name, structural_equation, indicators, structural_sigma=None)

Bases: object

Pure latent-variable specification.

Parameters:

• name (str) – Latent variable name.

• structural_equation (StructuralEquation) – Structural equation specification.

• indicators (Iterable[str]) – Names of indicators linked to this latent variable.

• structural_sigma (PositiveParameterSpec | None) – Optional specification of the structural standard deviation on its natural scale.

indicators: Iterable[str]

name: str

structural_equation: StructuralEquation

structural_sigma: PositiveParameterSpec | None

class biogeme.latent_variables.model_spec.LikertIndicator(name, statement, type_name)

Bases: object

Pure semantic indicator specification.

Parameters:

• name (str) – Indicator name.

• statement (str) – Human-readable text of the indicator.

• type_name (str) – Shared type metadata name.

name: str

statement: str

type_name: str

class biogeme.latent_variables.model_spec.LikertType(type_name, symmetric, categories, neutral_labels)

Bases: object

Shared indicator-type metadata.

Parameters:

• type_name (str) – Type name.

• symmetric (bool) – Whether the threshold system is symmetric when used by ordinal indicators.

• categories (list[int]) – Ordered category labels.

• neutral_labels (list[int]) – Neutral or placeholder labels.

categories: list[int]

neutral_labels: list[int]

symmetric: bool

type_name: str

class biogeme.latent_variables.model_spec.MeasurementConfiguration(specifications)

Bases: object

Collection of indicator-level measurement specifications.

Parameters:

specifications (Iterable[IndicatorMeasurementSpec]) – Measurement specification for each modeled indicator.

specifications: Iterable[IndicatorMeasurementSpec]

class biogeme.latent_variables.model_spec.MeasurementModel(*values)

Bases: str, Enum

Measurement model used for one indicator.

GAUSSIAN = 'gaussian'

ORDERED_LOGIT = 'ordered_logit'

ORDERED_PROBIT = 'ordered_probit'

class biogeme.latent_variables.model_spec.PositiveParameterSpec(start=None, lower_bound=0.0)

Bases: object

Specification of a strictly positive model parameter.

The value is expressed on the natural scale of the parameter. Downstream generators may transform it internally, for example by using log for a positive reparameterization.

Parameters:

• start (float | None) – Optional starting value on the natural scale.

• lower_bound (float | None) – Optional lower bound on the natural scale.

lower_bound: float | None

start: float | None

class biogeme.latent_variables.model_spec.StructuralEquation(name, intercept=True, explanatory_variables=())

Bases: object

Pure structural-equation specification.

Parameters:

• name (str) – Name of the owning latent variable.

• intercept (bool) – Whether a structural intercept should be included.

• explanatory_variables (Iterable[str]) – Variables entering the deterministic part.

explanatory_variables: Iterable[str]

intercept: bool

name: str