Dynamic Count Mixture Models (DCMM)

A review and discussion of paper by Berry & West (2019)

October 28, 2020

Berry and West (2019) proposed dynamic count mixture models aiming at forecasting count-valued time series containing multiple time granularities across multiple outlets. Here I summarise the decouple/recouple idea of the model and provide a brief introduction of their DCMM models and extensions.

Posted on:
October 28, 2020
1 minute read, 42 words
Bayesian Time series
See Also: