A0440
Title: Challenges in multivariate zero-inflated models with solutions using a copula approach
Authors: Anna van Es - University of Geneva (UNIGE) (Switzerland) [presenting]
Eva Cantoni - University of Geneva (Switzerland)
Abstract: Existing models for handling multivariate excess zeros typically focus on scenarios characterized by positive dependence among variables, which limits their applicability in more complex settings. The copula multivariate zero-inflated count model (MuZIC) is introduced, a novel approach that extends the current framework to accommodate a broader range of dependency structures - particularly in cases where the counts are low, and the Gaussian approximation is inappropriate. The flexibility and comprehensiveness of the model offer significant advantages over existing methods, as demonstrated through theoretical comparisons and extensive simulation studies. Finally, the computational challenges arising in higher dimensions are addressed by proposing potential solutions to improve scalability.
