Abstract: The COVID-19 pandemic continues to disrupt all areas of healthcare, economy and broader society. Individual models have shown limited predictive power in the dynamic environment and society within which COVID-19 spreads. Our first key objective was to identify some of the publicly available models and compile a repository of them.The second key objective was to create an Ensemble COVID-19 model which combines the predictive power of individual models and is able to more accurately forecast the COVID-19 cumulative cases in South Africa. We built two Ensemble models, namely, 1) Arithmetic Mean Ensemble 2)Weighted Average Ensemble. The Ensemble models outperformed the individual models by a significant margin. Furthermore, the Weighted Average Ensemble was better performing compared to the Arithmetic Mean Ensemble.

Presentation Video