Overview

The package implements the hierarchical Dirichlet process Generalized Linear Models proposed in the paper Modeling Context-Dependent Latent Effect Heterogeneity. The model can be used to estimate latent heterogeneity in the marginal effect of GLM linear coefficients, cluster data points based on that latent heterogeneity, and investigate if Simpson’s Paradox occurs due to latent or omitted features. It also can be used with hierarchical data to estimate the effect of upper-level covariates on the latent heterogeneity in the effect of lower-level features.

For details of the model and the MCMC algorithm, see Ferrari, D. (2020) Modeling Context-Dependent Latent Effect Heterogeneity, Political Analysis, Volume 28, Issue 1, January, pp. 20-46. <DOI: 10.1017/pan.2019.13>.

Instalation

From CRAN

install.packages("hdpGLM")

To install the development version, use:


devtools::install_github("DiogoFerrari/hdpGLM")

# If you don't want to update the dependencies, use: (you may need to install some dependencies manually)
devtools::install_github("DiogoFerrari/hdpGLM", dependencies=F)

NOTE: it may be necessary to create a token to install the package from the git repository in case it is private (see note at the bottom of help page in R by running help(install_github)).