In greta-dev/greta: Simple and Scalable Statistical Modelling in R

Multiple linear regression with LASSO prior

A multi-variable Bayesian linear regression model using an exponential-normal prior for the coefficients.

data(attitude)
design <- as.matrix(attitude[, 2:7])

int <- normal(0, 10)
sd <- cauchy(0, 3, truncation = c(0, Inf))

tau <- exponential(0.5, dim = ncol(design)) 
coefs <- normal(0, tau)
mu <- int + design %*% coefs

distribution(attitude$rating) <- normal(mu, sd)

greta-dev/greta documentation built on Dec. 21, 2024, 5:03 a.m.