Abstract:
In this research, the Poisson regression model is discussed in the case of a data set containing extreme values, and ways to overcome this problem were discussed using two robust methods (the weighted maximum likelihood estimation method and the robust mallows quasi-maximum likelihood estimation method) and a comparison between the two estimation methods to overcome the problem of extreme values and to know which is better. A sample size of (100) people was chosen and (6) tests were taken that diagnose Lupus disease or lupus erythematosus for the year (2022), as this real data was collected from private laboratories affiliated with the Medical City and the Iraqi Ministry of Health. It was shown through the results that the extreme values were treated through the two robust methods, and the best method was the robust mallows quasi-maximum likelihood estimation method.
Keywords : Generalized linear models , Poisson regression , Outliers , Robust regression methods , Robust mallows quasi-maximum likelihood (MQL) , Weighted maximum likelihood (WML) .
