Generalized Linear Model analysis for Binomial distribution with the application

Abstract

The aim of this research is to use a generalized Linear model for the binary logistic regression model to study the effect of different concentrations of two drugs (xi1, xi2) that were studied at different levels on patients with nephritis (kidney inflammation), where the response variable (yi) represents the number of cure cases (Binomial distribution) as a result of taking the two drugs. In order to estimate the logistic model, two functions link were used, the first is the probit function, and the second is the logit function, and then a comparison between the results of using the two functions. The results support the preference of the estimated Logit regression model through the criteria (Akaike's Information Criterion, Bayesian Information Criterion, and Mean Square Error). The effectiveness of the first independent variable (drug) on the second independent variable (drug) through the preferred model.