Forecasting count data using time series model with exponentially decaying covariance structure
share
Count data appears in various disciplines. In this work, a new method to analyze time series count data has been proposed. The method assumes exponentially decaying covariance structure, a special class of the Matérn covariance function, for the latent variable in a Poisson regression model. It is implemented in a Bayesian framework, and can provide reliable estimates for covariate effects and extent of variability explained by the temporally dependent process and the white noise process. Prediction procedure has been described as well. Simulation study with four different processes show that across various scenarios the newly proposed method generally has better predictive accuracy than other popular methods. Further, two real data examples, one related to the number of dengue cases and another on the number of sales made by a retailer, are included in the paper. These two examples corroborate the earlier findings and establish that the proposed approach has good predictive abilities.
READ FULL TEXT
