A hybrid simple exponential smoothing-barnacles mating optimization approach for parameter estimation: Enhancing COVID-19 forecasting in Malaysia

Single or simple exponential smoothing (SES) is a time series forecasting model popular among researchers due to its simplicity and ease of use. SES only requires one smoothing parameter, alpha, to control how quickly the influence of past observations decreases. However, SES is seen to underperform compared to other models due to parameter selection and initial value setting. Therefore, this study aims to propose a new hybrid model, the Single Exponential Smoothing (SES)-Barnacles Mating Optimization (BMO) algorithm, to estimate the optimal smoothing parameter alpha and initial value that can improve the percentage of forecast accuracy. Some of the highlights of the proposed method are:
• A new hybrid model, SES-BMO, has successfully estimated the optimal initial value and smoothing parameter simultaneously with a high forecast accuracy (90.2 %).
• The data splitting ratio 80:20 or 75:25 is unsuitable for research cases requiring immediate action and decision, such as the COVID-19 pandemic. Thus, implementing Repeated timeseries cross-validation (RTS-CV) is a good practice in model validation.
• The average 8-day forecast accuracy is 90.2 %. The lowest and highest forecast accuracy was 83.7 % and 98.8 %.