Bayesian Forecasting of Many Count-Valued Time Series

Journal Article (Journal Article)

We develop and exemplify application of new classes of dynamic models for time series of nonnegative counts. Our novel univariate models combine dynamic generalized linear models for binary and conditionally Poisson time series, with dynamic random effects for over-dispersion. These models estimate dynamic regression coefficients in both binary and nonzero count components. Sequential Bayesian analysis allows fast, parallel analysis of sets of decoupled time series. New multivariate models then enable information sharing in contexts when data at a more highly aggregated level provide more incisive inferences on shared patterns such as trends and seasonality. A novel multiscale approach—one new example of the concept of decouple/recouple in time series—enables information sharing across series. This incorporates cross-series linkages while insulating parallel estimation of univariate models, and hence enables scalability in the number of series. The major motivating context is supermarket sales forecasting. Detailed examples drawn from a case study in multistep forecasting of sales of a number of related items showcase forecasting of multiple series, with discussion of forecast accuracy metrics, comparisons with existing methods, and broader questions of probabilistic forecast assessment.

Full Text

Duke Authors

Cited Authors

  • Berry, LR; West, M

Published Date

  • October 1, 2020

Published In

Volume / Issue

  • 38 / 4

Start / End Page

  • 872 - 887

Electronic International Standard Serial Number (EISSN)

  • 1537-2707

International Standard Serial Number (ISSN)

  • 0735-0015

Digital Object Identifier (DOI)

  • 10.1080/07350015.2019.1604372

Citation Source

  • Scopus