Behavioral analysis of COVID-19 daily deaths in Sri Lanka

Abstract

In December 2019, the world was unknowingly caught by coronavirus disease. Since it spread worldwide and the burden of this disease is very high, the whole world has tried to reduce the impact. Therefore, known scientific facts about future events are very useful for policymakers. Hence, this research contributes to a scientific and technological process to provide a method for controlling the movements of the general public by forecasting the future using Discrete Fourier Transform (DFT) technique. DFT technique can decompose a complex signal into simpler parts to facilitate the analysis. In this research, comparing the accuracy of the daily deaths and daily infected patient datasets, the daily deaths dataset was selected for the model. After that, the Sri Lankan COVID-19 daily deaths data subsequence was modelled as the DFT amplitude spectrum. Then by backward-transformation of significant harmonics predicts the near future of daily COVID-19 deaths. The results conclude that the proposed method can provide short-term forecasting of COVID-19 daily death data with sufficient accuracy. And these near-future predictions will help guide health planning and short-term policy changes.