Model Selection in Multivariate Longitudinal Data Analysis

An efficient Gibbs sampling algorithm is developed for Bayesian multivariate longitudinal data analysis with the focus on selection of important elements in the generalized autoregressive matrix. It provides posterior samples and estimates of parameters. In addition, estimates of several information criteria such as Akaike information criterion (AIC), Bayesian information criterion (BIC), deviance information criterion (DIC) and prediction accuracy such as the marginal predictive likelihood (MPL) and the mean squared prediction error (MSPE) are provided for model selection.

Utilising Normalisation Constant Optimisation via Edge Removal (UNCOVER)

Model data with a suspected clustering structure (either in co-variate space, regression space or both) using a Bayesian product model with a logistic regression likelihood. Observations are represented graphically and clusters are formed through various edge removals or additions. Cluster quality is assessed through the log Bayesian evidence of the overall model, which is estimated using either a Sequential Monte Carlo sampler or a suitable transformation of the Bayesian Information Criterion as a fast approximation of the former. The internal Iterated Batch Importance Sampling scheme (Chopin (2002 )) is made available as a free standing function.

Contrast-Based Bayesian Network Meta Analysis

A function that facilitates fitting three types of models for contrast-based Bayesian Network Meta Analysis. The first model is that which is described in Lu and Ades (2006) . The other two models are based on a Bayesian nonparametric methods that permit ties when comparing treatment or for a treatment effect to be exactly equal to zero. In addition to the model fits, the package provides a summary of the interplay between treatment effects based on the procedure described in Barrientos, Page, and Lin (2023) .

Bayesian Penalized Quantile Regression

Bayesian regularized quantile regression utilizing two major classes of shrinkage priors (the spike-and-slab priors and the horseshoe family of priors) leads to efficient Bayesian shrinkage estimation, variable selection and valid statistical inference. In this package, we have implemented robust Bayesian variable selection with spike-and-slab priors under high-dimensional linear regression models (Fan et al. (2024) and Ren et al. (2023) ), and regularized quantile varying coefficient models (Zhou et al.(2023) ). In particular, valid robust Bayesian inferences under both models in the presence of heavy-tailed errors can be validated on finite samples. Additional models with spike-and-slab priors include robust Bayesian group LASSO and robust binary Bayesian LASSO (Fan and Wu (2025) ). Besides, robust sparse Bayesian regression with the horseshoe family of (horseshoe, horseshoe+ and regularized horseshoe) priors has also been implemented and yielded valid inference results under heavy-tailed model errors(Fan et al.(2025) ). The Markov chain Monte Carlo (MCMC) algorithms of the proposed and alternative models are implemented in C++.

Additive Model for Ordinal Data using Laplace P-Splines

Additive proportional odds model for ordinal data using Laplace P-splines. The combination of Laplace approximations and P-splines enable fast and flexible inference in a Bayesian framework. Specific approximations are proposed to account for the asymmetry in the marginal posterior distributions of non-penalized parameters. For more details, see Lambert and Gressani (2023) ; Preprint: ).

Multivariate (Dynamic) Generalized Additive Models

Fit Bayesian Dynamic Generalized Additive Models to multivariate observations. Users can build nonlinear State-Space models that can incorporate semiparametric effects in observation and process components, using a wide range of observation families. Estimation is performed using Markov Chain Monte Carlo with Hamiltonian Monte Carlo in the software 'Stan'. References: Clark & Wells (2023) .

A Shiny Application for End-to-End Bayesian Decision Network Analysis and Web-Deployment

A Shiny application for learning Bayesian Decision Networks from data. This package can be used for probabilistic reasoning (in the observational setting), causal inference (in the presence of interventions) and learning policy decisions (in Decision Network setting). Functionalities include end-to-end implementations for data-preprocessing, structure-learning, exact inference, approximate inference, extending the learned structure to Decision Networks and policy optimization using statistically rigorous methods such as bootstraps, resampling, ensemble-averaging and cross-validation. In addition to Bayesian Decision Networks, it also features correlation networks, community-detection, graph visualizations, graph exports and web-deployment of the learned models as Shiny dashboards.

Bayesian Distributed Lag Model Fitting for Binary and Count Response Data

Tools for fitting Bayesian Distributed Lag Models (DLMs) to longitudinal response data that is a count or binary. Count data is fit using negative binomial regression and binary is fit using quantile regression. The contribution of the lags are fit via b-splines. In addition, infers the predictor inclusion uncertainty. Multimomial models are not supported. Based on Dempsey and Wyse (2025) .

Reference Analysis for Bayesian Meta-Analysis

Functionality for performing a principled reference analysis in the Bayesian normal-normal hierarchical model used for Bayesian meta-analysis, as described in Ott, Plummer and Roos (2021) . Computes a reference posterior, induced by a minimally informative improper reference prior for the between-study (heterogeneity) standard deviation. Determines additional proper anti-conservative (and conservative) prior benchmarks. Includes functions for reference analyses at both the posterior and the prior level, which, given the data, quantify the informativeness of a heterogeneity prior of interest relative to the minimally informative reference prior and the proper prior benchmarks. The functions operate on data sets which are compatible with the 'bayesmeta' package.

Sparse Generalized Additive Models

Fits sparse frequentist GAMs (SF-GAM) for continuous and discrete responses in the exponential dispersion family with the group lasso, group smoothly clipped absolute deviation (SCAD), and group minimax concave (MCP) penalties . Also fits sparse Bayesian generalized additive models (SB-GAM) with the spike-and-slab group lasso (SSGL) penalty of Bai et al. (2021) . B-spline basis functions are used to model the sparse additive functions. Stand-alone functions for group-regularized negative binomial regression, group-regularized gamma regression, and group-regularized regression in the exponential dispersion family with the SSGL penalty are also provided.