Abstract

Regression models for counts could be applied to the earth sciences, for instance when studying trends of extremes of climatological quantities. Hurdle models are modified count models which can be regarded as mixtures of distributions. In this paper, hurdle models are applied to model the sums of lengths of periods of high temperatures. A modification to the common versions presented in the literature is presented, as left truncation as well as a particular treatment of zeros is needed for the problem. The outcome of the model is compared to those of simpler count models.

Keywords

Count data; hurdle models; Poisson regression; negative binomial distribution; climate

Published in

SORT
2017, Volume: 41, number: 1, pages: 177-188

UKÄ Subject classification

Probability Theory and Statistics


Publication identifier

DOI: https://doi.org/10.2436/20.8080.02.57

Permanent link to this page (URI)

https://res.slu.se/id/publ/99688