Comparıson Study of The Poisson Regression Model Parameters Estımated With Different tow Methods(statistical study)

The aim of this study is comparative examination of the estimation methods where can be employed to estimate Poisson regression model parameters. Occurrence number of any events that takes place within a specified time period as a result of conducted experiments can be expressed as count data. Poisson regression model is employed as an important data interpretation tool to analyze this kind of count data. Poisson regression models are regarded as a sub-branch of generalized linear models. The following tow methods are used for parameters estimation: 1(Maximum Likelihood Estimation (MLE),2) linear least squares(OLS). MATLAB packaged software is used for generation of simulation data and for parameter estimates. Poisson regression model parameters were estimated and models were generated by using of Monte Carlo simulation with sample sizes of 30, 60, 90 and 100 in accordance with Poisson distribution. Mean square error (MSE) and mean absolute percentage error (MAPE) criteria were used for comparison of estimated parameters in terms of their effectiveness Mean square error (MSE) and mean absolute percentage error (MAPE) criteria were used for comparison of estimated parameters in terms of their effectiveness. As a result of comparison, it was shown that MLE gives better results than other method OLS.