Sliding window time series forecasting with multilayer perceptron and multiregression of COVID-19 outbreak in Malaysia

This study demonstrates a sliding window time series forecasting methods to predict future trends of pandemic coronavirus disease 2019 (COVID-19) reported in Malaysia using a multiple regression and single-layer feedforward artificial neural network. Data from Jan. 25 to Apr. 30 were obtained from the Malaysian Ministry of Health and Department of Statistics Malaysia website. The findings show that the Movement Control Order declared by the Malaysian government was effective in mitigating the risk for spreading COVID-19 diseases through home quarantine and isolation, and thus were able to flatten the curve. Sliding window time series forecasting with an artificial neural network performs better than multiple regression as a predictive model with a smaller residual error.