A Flexible Count Regression Model Based On Poisson-Juchez Distribution

Abstract

A new count regression model known as Poisson-Juchez regression model is proposed in this study. The model is based on the Poisson-Juchez Distribution and is aimed to serve as a suitable count regression model. Its theoretical framework, estimation technique and simulation results are being studied. The proposed Poison-Juchez regression model is being compared to traditional Poisson regression and Negative Binomial regression models to demonstrate its flexibility and applicability to real life data. The model selection criteria result for the dataset showed that Poisson-Juchez regression model achieves the best Akaike Information Criterion AIC (200.81) and Bayesian Information Criterion BIC (210.54), thus outperforming both the standard Poisson (AIC = 203.26, BIC = 212.98) and Negative Binomial (AIC = 202.79, BIC =215.75). This implies that the Poisson-Juchez regression model is the preferred model, as it had the least AIC and BIC. The finding from this study had substantiated that pre-pregnancy weight (lwt) is a significant predictor of prior preterm labour, and this is being upheld across all three models, thus establishing it as the key determinant of prior preterm labour frequency in this study.