Approximately linear INGARCH models for spatio-temporal counts
针对现有时空计数INGARCH模型不允许负参数和自相关的局限，提出统一的近似线性INGARCH框架，可灵活处理有界和无界计数，并应用于降水与云量数据。
Abstract Existing integer-valued generalised autoregressive conditional heteroskedasticity (INGARCH) models for spatio-temporal counts do not allow for negative parameter and autocorrelation values. Using approximately linear INGARCH models, the unified and flexible spatio-temporal (B)INGARCH framework for modelling unbounded (bounded) counts is proposed. These models combine negative dependencies with kinds of a long memory. They are easily adapted to special marginal features or cross-dependencies: When modelling precipitation data (counts of rainy hours), we account for zero-inflation, while for cloud-coverage data (counts of okta), we deal with missing data and additional cross-correlation. A copula related to the spatial error model shows an appealing performance.