elizabethchase/HPR: Horseshoe Process Regression

HPR performs horseshoe process regression, as described in the article by Chase et al. (2022+). HPR is an analogue to Gaussian process regression (GPR), but intended for more abruptly-varying functions, like step functions, than can be modeled by GPR. It allows for normal, Poisson, and binomial outcome data, with a mixture of HPR and linear predictors, which can be constrained to be monotonic increasing or decreasing. For more information, see Chase et al. (2022+) and the vignette.

install.packages("remotes")
remotes::install_github("elizabethchase/HPR")