Liu Estimator in the Multinomial Logistic Regression Model

TL;DR

This paper investigates the use of Liu estimators in multinomial logistic regression, proposing new biasing parameter estimators and evaluating their performance through simulation to improve model accuracy.

Contribution

It introduces new estimators for the biasing parameter in multinomial logistic regression and compares their performance using Monte Carlo simulations.

Findings

Higher correlation increases MSE.

Larger sample size reduces MSE.

Certain estimators outperform others based on MSE.

Abstract

This paper considers the Liu estimator in the multinomial logistic regression model. We propose some different estimators of the biasing parameter. The mean square error (MSE) is considered as the performance criterion. In order to compare the performance of the estimators, we performed a Monte Carlo simulation study. According to the results of the simulation study, we found that increasing the correlation between the independent variables and the number of regressors has a negative effect on the MSE. However, when the sample size increases the MSE decreases even when the correlation between the independent variables is large. Based on the minimum MSE criterion some useful estimators for estimating the biasing parameter d are recommended for the practitioners.