Stata 15 help for fmm_estimation

[FMM] fmm estimation -- Fitting finite mixture models


Finite mixture models (FMMs) are used to classify observations, to adjust for clustering, and to model unobserved heterogeneity. In finite mixture modeling, the observed data are assumed to belong to several unobserved subpopulations called classes, and mixtures of probability densities or regression models are used to model the outcome of interest. After fitting the model, class membership probabilities can also be predicted for each observation.

Linear regression models

[FMM] fmm: regress Linear regression [FMM] fmm: truncreg Truncated regression [FMM] fmm: intreg Interval regression [FMM] fmm: tobit Tobit regression [FMM] fmm: ivregress Instrumental-variables regression

Binary-response regression models

[FMM] fmm: logit Logistic regression, reporting coefficients [FMM] fmm: probit Probit regression [FMM] fmm: cloglog Complementary log-log regression

Ordinal-response regression models

[FMM] fmm: ologit Ordered logistic regression [FMM] fmm: oprobit Ordered probit regression

Categorical-response regression models

[FMM] fmm: mlogit Multinomial (polytomous) logistic regression

Count-response regression models

[FMM] fmm: poisson Poisson regression [FMM] fmm: nbreg Negative binomial regression [FMM] fmm: tpoisson Truncated Poisson regression

Generalized linear models

[FMM] fmm: glm Generalized linear models

Fractional-response regression models

[FMM] fmm: betareg Beta regression

Survival regression models

[FMM] fmm: streg Parametric survival models

fmm: allows different regression models for different components of the mixture; see [FMM] fmm. fmm: also allows one or more components to be a degenerate distribution taking on a single integer value with probability one; see [FMM] fmm: pointmass.

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index