Predicting COVID-19 Induced Mortality Utilizing Vaccination Status Data

Abstract

COVID-19 is a prevalent pandemic that has caused of millions of mortalities. While medical and statistical research have proven the effectiveness of COVID vaccines to reduce mortality rate in specific areas. This study aims to explore the quantitative effect of vaccine in preventing in the global level. Mortality and vaccine data was acquired from Kaggle, which summarized related data from various sources. Multiple linear regression, quasi-Poisson regression, LASSO regression, and random forest are applied to the model to analyze and predict the effect of vaccine on mortality. The adjusted R square value of these four models is 0.6670, 0.6863, 0.4901, and 0.7152 respectively. The residual boxplots of all four models show that LASSO model sometimes make extremely inaccurate predictions, and the random forest model is the most accurate model, which matches the comparison result of the adjusted R square values. Therefore, the random forest model can be potentially used to predict future COVID-related deaths based on vaccine-related data.