drkowal/fosr: Function-on-Scalars Regression for High Dimensional Data

Bayesian inference for function-on-scalars regression, where a functional data response is regressed on scalar predictors. The functions are modeled nonparametrically using an unknown basis expansion, which is learned from the data. The regression coefficients themselves are functions, and are assigned shrinkage priors to reduce the impact of unimportant predictor variables. Posterior inference is available via an efficient MCMC algorithm, which is valid and scalable for p > n. For Bayesian variable selection in functional regression, we include two approaches: (i) a decision-theoertic approach that selects variables jointly and (ii) a Global Bayesian P-value approach based on simulataneous credible bands that selects variables marginally.

install.packages("remotes")
remotes::install_github("drkowal/fosr")