Pemodelan Regresi dengan Pendekatan Generalized Linear Models pada Kasus Tuberkulosis di Sumatera Utara

Abstract

Generalized Linear Models (GLM) are used to model data when the distribution of the response variable is part of the exponential family. The negative binomial distribution and the Poisson distribution are examples of exponential distributions. While the link function explicitly explais the relationship between the regression model and the expected value of the response variable, the GLM model explains the structure of the predictor variables. The aim of this study is to identify predictor variables that significantly influence the optimal model. The incidence of tuberculosis in North Sumatra will be modeled using Poisson and negative binomial regression. Based on the results of the analysis, it was found that two predictor variables significantly influenced the incidence of tuberculosis in North Sumatra, namely the number of poor people and the percentage of people aged 45-54 years who smoke, with AIC value =483,91. where the model chosen is negative binomial regression with the model equation π=exp(4,861+0,000021x1+0,025610x3).