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

Multiple categorical regression

A multi-variable Bayesian categorical regression model using the iris data.

data(iris)
X <- as_data(cbind(1, iris[, 1:4]))
y <- model.matrix(~ Species - 1, iris)
P <- ncol(X)
K <- ncol(y)

beta <- normal(0, 5, dim = c(P, K - 1))
eta <- X %*% beta
prob <- imultilogit(eta)
distribution(y) <- categorical(prob)

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