Repository navigation
Releases
TDMM v0.3.0
TDMM v0.3.0 adds support for time-dynamic regression models without a subject-specific random intercept.
- Set
random_intercept = FALSEintdmm()for independent observations. The default remainsTRUEfor repeated observations, preserving existing workflows. - Gaussian, Bernoulli, and Poisson simulation generators now support both repeated- and independent-observation settings.
- The Getting Started guide, family examples, and reference documentation show how to use both settings.
TDMM v0.2.0
TDMM v0.2.0 supports Gaussian, Bernoulli, and Poisson longitudinal outcomes with subject-specific irregular observation times, unequal visit counts, baseline covariates, and time-varying covariates. New fits use cubic B-splines with a second-order coefficient-difference penalty and eight basis functions by default. The number of basis functions is controlled directly with n_basis.
The recommended workflow is tdmm() followed by summary(), coefficient plots, and chain diagnostics. Serial and parallel fitting are selected with tdmm_control(); tdmm.parallel() remains available for compatibility. The package includes a Getting Started vignette, short examples for all three families, and a reproducible PBC2 application. Users should assess convergence and model fit for each analysis.
Version 0.1.0 remains preserved as the historical V1 release. The Zenodo DOI 10.5281/zenodo.20617705 corresponds to v0.1.0. TDMM v0.2.0 is archived at DOI 10.5281/zenodo.23608508.
TDMM v0.1.0
Initial release of the TDMM R package for Bayesian time-dynamic mixed-effects models for longitudinal Gaussian, Bernoulli, and Poisson outcomes.
This release includes data checking functions, model fitting through JAGS, posterior summaries, MCMC diagnostic tools, visualization functions, and tutorial documentation.