Title: Zero-inflated exponential families, with applications to count times-series
Authors: Alan Huang - University of Queensland (Australia) [presenting]
Thomas Fung - Macquarie University (Australia)
Abstract: A novel mechanism is introduced that generates exponential families of zero-inflated distributions from a given nominal distribution. This proves particularly useful for observation-driven times-series modelling of zero-inflated counts, where neither the underlying zero-inflation nor count processes are fully observable. The idea is demonstrated on two data analysis examples.