A Comparison Study on Distributional Assumption Tests for Poisson Regression Model

• Activity Number: 702
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Deniz Ozonur* and Hatice Tul Kubra Akdur and Hulya Bayrak
• Companies: Gazi University and Gazi University and Gazi University
• Keywords: Poisson regression model ; Goodness of fit ; Type I error ; Power of test

Abstract

Poisson regression model is appropriate when response variable takes discrete values however, it is not a categorical variable. For example, the number of insects in a leaf or the number of fires occurring in a year and so forth. Poisson regression model is a special case of generalized linear models with Poisson error and a log link. It is firstly described by Nelder and Wedderburn (1972). Poisson and Logistic regression models are investigated by using smooth tests by Rippon and Rayner (2011). Moreover, Ye et al. (2013) used this model to analyze relationship between accidents and reasons and they researched the suitability of the model.Goodness of fit tests how well a set of observations fits to a statistical model. In this study goodness of fit tests such as Power- Divergence, Freeman-Tukey, Pearson chi-squared, Deviance tests, smooth tests and components, Dean and Lawless's score test are examined for Poisson regression model. These tests are compared in terms of type I error rates and power values by simulation study.