Multivariate Count Data Models for Time Series Forecasting

Yuliya Shapovalova*, Nalan Basturk, Michael Eichler

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Count data appears in many research fields and exhibits certain features that make modeling difficult. Most popular approaches to modeling count data can be classified into observation and parameter-driven models. In this paper, we review two models from these classes: the log-linear multivariate conditional intensity model (also referred to as an integer-valued generalized autoregressive conditional heteroskedastic model) and the non-linear state-space model for count data. We compare these models in terms of forecasting performance on simulated data and two real datasets. In simulations, we consider the case of model misspecification. We find that both models have advantages in different situations, and we discuss the pros and cons of inference for both models in detail.
Original languageEnglish
Article number718
Number of pages23
Issue number6
Publication statusPublished - Jun 2021


  • multivariate count data
  • state-space model
  • bank failures
  • transactions


Dive into the research topics of 'Multivariate Count Data Models for Time Series Forecasting'. Together they form a unique fingerprint.

Cite this