LL_cumulative: Proportional odds model log-likelihood

In gspcr: Generalized Supervised Principal Component Regression

Proportional odds model log-likelihood

Description

Computes the log-likelihood given an ordered category response vector and corresponding GLM linear predictor values.

Usage

LL_cumulative(y, x, mod)

Arguments

y	ordered factor or disjunctive table representation recording an ordinal variable with 3 or more categories.
x	data.frame (or matrix) containing predictor values.
mod	polr object containing the estimated proportional odds model.

Details

If x and y are equal to the data on which mod has been trained, this function returns the same result as the default logLink function. If x and y are new, the function returns the log-likelihood of the new data under the trained model. The log-likelihood equation is based on Agresti (2002, p. 192).

Value

A list containing:

• ll an atomic vector of length 1 containing the log-likelihood value.

• sc, a numeric matrix containing the systematic component for the input x and mod.

Author(s)

Edoardo Costantini, 2022

References

Agresti, A. (2012). Categorical data analysis (Vol. 792). John Wiley & Sons.

gspcr documentation built on May 29, 2024, 2:44 a.m.