Advances and Applications in Statistics
Volume 3, Issue 2, Pages 145 - 158
(August 2003)
|
|
ON INFLATED GENERALIZED POISSON REGRESSION MODELS
Felix Famoye (U. S. A.) and Karan P. Singh (U. S. A.)
|
Abstract: Zero-inflated
Poisson and zero-inflated negative binomial
regression models have been proposed for
datasets that result into too many zeros.
Recently, the zero-inflated generalized Poisson
regression model is defined as a good alternate
to model count data with too many zeros [Famoye
and Singh, Zero inflated generalized Poisson
regression model, submitted]. In this paper, we
propose a k-inflated generalized Poisson
regression (k-IGPR) to model count data
with too many k-values. Estimation of the
model parameters using the method of maximum
likelihood is provided. A score test is
presented to test whether the number of k-values
is too large for the generalized Poisson model
to adequately fit the data. The k-inflated
generalized Poisson regression model is
illustrated using a dataset with too many ones. |
Keywords and phrases: count data, k-inflation, estimation, hypothesis testing. |
|
Number of Downloads: 348 | Number of Views: 1242 |
|