A multivariate Poisson regression model for count data

Area of Science:

• Ecology
• Paleontology
• Statistical Modeling

Background:

• Multivariate count data presents analytical challenges, particularly in ecological and paleontological studies.
• Existing models may not adequately capture the complex correlations present in species abundance data across geographical locations.

Purpose of the Study:

• To develop and validate a novel statistical technique for the analysis of multivariate count data.
• To apply this technique to model the abundance of fossil species across various geographical observation points.

Main Methods:

• A multivariate model based on Poisson distributions is proposed, allowing for both positive and negative correlations.
• The log-linear Poisson model is extended to the multivariate case using conditional distributions.
• Maximum likelihood estimates and goodness-of-fit statistics are derived and computed.

Main Results:

• The proposed method is demonstrated on simulated datasets, validating its performance.
• The technique is successfully applied to a real-world dataset of fossil species counts.

Conclusions:

• The new multivariate Poisson model provides a robust framework for analyzing complex count data.
• This method enhances the study of species distribution and abundance in ecological and paleontological research.