A New Count Regression Model including Gauss Hypergeometric Function with an application to model demand of health services
share
In this paper, an alternative count distribution suitable for modeling over dispersed, zero vertex unimodality and monotonically decreasing data sets. Though the proposed probability model includes Gauss Hypergeometric special function, it possesses simple and closed expressions for various distributional characteristics. An application to count regression modeling using a well-known data set from the National Medical Expenditure Survey is discussed by considering the length of stay in hospitalization as a dependent variable and following the proposed count model. We compare our result with the classical Negative Binomial regression model and recently proposed Uniform Poisson regression model.
READ FULL TEXT