Parameters Estimation of Multinomial Probit Regression Model Using Maximum Likelihood Estimation

Abstract

Multinomial probit regression (MNP) is a regression analysis that identifies the relationship between the dependent variable and one or more independent variables that are categorical, where the model used is a logistic model whose link function uses the inverse of the normal distribution cumulative function. To form an MNP model, parameter estimation is carried out to measure the relative influence of each independent category on the dependent category in the MNP model. The method used to estimate the MNP model is maximum likelihood estimation (MLE). The test results by taking a case example of Maternal Health Risk show that the multinomial probit model whose parameters have been estimated using MLE can provide a model accuracy rate of 67.88% where the model for