biogeme.latent_variables.measurement_equations module¶
JAX-compatible ordered-probit measurement equations for Biogeme latent-variable models.
This module constructs the joint ordered-probit measurement model typically used in hybrid choice models. It combines:
latent-variable definitions (structural equations, normalization rules),
ordinal indicators (intercepts, latent-variable loadings, measurement scales),
Likert-scale definitions (categories and ordered thresholds),
consistency checks across model components,
JAX-friendly Biogeme
Expressionobjects.
Thresholds (cut-points) are obtained from the LikertType
associated with each indicator via get_thresholds().
Threshold sharing relies on parameter naming: if multiple indicators (or types)
produce cut-points with identical parameter names, Biogeme will estimate a
single shared set of parameters.
The main entry points are:
measurement_equations_jax(): product of measurement likelihoods.log_measurement_equations_jax(): sum of log measurement likelihoods.
Michel Bierlaire Thu Dec 11 2025, 16:26:07
- biogeme.latent_variables.measurement_equations.log_measurement_equations_jax(latent_variables, likert_indicators, likert_types, draw_type)[source]¶
Return the sum of log ordered-probit measurement likelihood terms.
- Parameters:
latent_variables (
list[LatentVariable]) – Latent-variable objects defining structural equations and associated indicators.likert_indicators (
list[LikertIndicator]) – Likert indicator objects providing measurement parameters.likert_types (
list[LikertType]) – Likert-type objects defining categories and thresholds.draw_type (
str) – Draw type assigned to latent variables for JAX.
- Return type:
- Returns:
Expression representing the sum of log ordered-probit likelihood terms.
- biogeme.latent_variables.measurement_equations.measurement_equations_jax(latent_variables, likert_indicators, likert_types, draw_type)[source]¶
Return the product of ordered-probit measurement likelihood terms.
- Parameters:
latent_variables (
list[LatentVariable]) – Latent-variable objects defining structural equations and associated indicators.likert_indicators (
list[LikertIndicator]) – Likert indicator objects providing measurement parameters.likert_types (
list[LikertType]) – Likert-type objects defining categories and thresholds.draw_type (
str) – Draw type assigned to latent variables for JAX.
- Return type:
- Returns:
Expression representing the product of ordered-probit likelihood terms.