predict: Predict method for MGLM Fits

In MGLM: Multivariate Response Generalized Linear Models

Predict method for MGLM Fits

Description

Predict using the fitted model from MGLMreg when given a new set of covariates.

Usage

## S4 method for signature 'MGLMreg'
predict(object, newdata)

Arguments

object	model object.
newdata	new covariates data matrix.

Value

Outputs the probabilities of each category.

This helps answer questions such as whether certain features increase the probability of observing category j.

Examples

n <- 200
p <- 5
d <- 4
X <- matrix(runif(p * n), n, p)
alpha <- matrix(c(0.6, 0.8, 1), p, d - 1, byrow=TRUE)
alpha[c(1, 2),] <- 0
Alpha <- exp(X %*% alpha) 
beta <- matrix(c(1.2, 1, 0.6), p, d - 1, byrow=TRUE)
beta[c(1, 2),] <- 0
Beta <- exp(X %*% beta)
m <- runif(n, min=0, max=25) + 25
Y <- rgdirmn(n, m, Alpha, Beta)
gdmReg <- MGLMreg(Y~0+X, dist="GDM")
newX <- matrix(runif(1*p), 1, p)
pred <- predict(gdmReg, newX)

MGLM documentation built on April 14, 2022, 1:07 a.m.