Variable Selection in Poisson Regression Model using Golden Jackal Optimization Algorithm

Abstract

The Poisson regression model is one of the most important logarithmic linear regression models, and it is the tool through which the response variable is modeled when the values ​​of that variable are in the form of countable values. Like other regression models, the model may contain many explantory variables, which negatively affects the accuracy of the model and its simplicity in interpreting the results. This study aims to use the Golden Jackal algorithm and compare it with other methods in selecting variables in the Poisson regression model using simulation and real data. The Monte-Carlo method was used in the simulation to generate data that follow the Poisson regression model according to different factors such as sample size and the number of explantory variables. Two aspects of the performance evaluation of the methods used were relied upon: the first is to evaluate the accuracy of prediction and the second is to evaluate the selection of variables as a criterion for comparison. The simulation results showed the superiority of the Golden Jackal algorithm compared to other variable selection methods. In addition, the application was carried out on real data collected from patients with chronic kidney disease who are treated with continuous hemodialysis, and the patients' condition was diagnosed by specialist doctors in cooperation with Ibn Sina Teaching Hospital - Artificial Kidney Unit.