The XBART-Poisson Classification Model for COVID-19 Data Analysis in Egypt

This paper aims to predict daily case, mortality counts, classify high-risk periods and provide interpretable, probabilistic insights into COVID-19 trends in Egypt by using Extreme Bayesian Additive Regression Trees with Poisson likelihood (XBART-Poisson) model to COVID-19 data in Egypt. The model is adapted for the pandemic's count-based data, such as daily cases, mortality counts, and recovery rates, offering a Bayesian probabilistic approach to forecast trends and analyze epidemiological factors. The Poisson likelihood effectively handles the discrete nature of these data points. Performance is benchmarked against traditional classification techniques, revealing XBART-Poisson’s robustness in capturing key trends and providing accurate predictions for COVID-19 progression in Egypt. The study reaches the suggested model which is more accurate than the traditional models such as Logistic Regression, Decision Tree, Random Forest and XGBoost.