Differentiation of various estimators in handling multicollinearity in poisson regression model: Case study of infant mortality rate in East Nusa Tenggara Province, Indonesia
Author(s): Fanny Winda Aini, Netti Herawati, Misgiyati and Nusyirwan
Abstract: Poisson regression is used to model Poisson-distributed count data as a dependent variable with one or more independent variables. Poisson regression uses the Maximum Likelihood Estimation method by considering the assumptions that must be met, including the absence of multicollinearity. The assumption of multicollinearity in Poisson regression can cause large variance in the data. In this study, Poisson regression with three different estimator methods will be compared, namely the Poisson James Stein Estimator, the Poisson Ridge Regression Estimator and the Poisson Kibria Lukman estimator, on infant mortality rates in East Nusa Tenggara Province, Indonesia, where there is a multicollinearity problem is based on the MSE value. The estimator with the best MSE will be declared as the best estimator. The results show the Poisson Ridge Regression Estimator method, with a parameter ridge of 0.00007, is the most effective in handling multicollinearity, because it produces the smallest Mean Squared Error.
Fanny Winda Aini, Netti Herawati, Misgiyati, Nusyirwan. Differentiation of various estimators in handling multicollinearity in poisson regression model: Case study of infant mortality rate in East Nusa Tenggara Province, Indonesia. Int J Multidiscip Trends 2025;7(3):40-45. DOI: 10.22271/multi.2025.v7.i3a.626
of 0.00007, is the most effective in handling multicollinearity, because it produces the smallest Mean Squared Error.
Research Journals