Estimasi Parameter Regresi Logistik Biner Menggunakan Metode Maximum Likelihood

Abstract

This study discusses parameter estimation in binary logistic regression using maximum likelihood. Binary logistic regression is a logistic regression with the dependent variable being dichotomous or consisting of 2 categories. The parameter estimation method used is maximum likelihood estimation. The model parameters were simultaneously tested by the likelihood ratio test and the model parameters were partially tested by the wald test. Logit transformation was performed to obtain a binary logistic regression model. Based on the results obtained, there are 3 independent variables that significantly affect chronic obstructive pulmonary disease, namely FEV1, FVC, and FEF 25-75. From the likelihood ratio test obtained a significance value of a≤0.05 so it can be concluded that the model is significant. From the Wald test the independent variables that affect chronic obstructive pulmonary disease are FEV1, FVC, and FEF 25-75. The logistic regression model obtained is g(x)=−0.488 + 0.083 x1− 0.182 x2+ 0.143 x3− 0.036 x4