Forecasting the Cumulative COVID-19 Cases in Indonesia Using Flower Pollination Algorithm

• Published in: Computation
• Main Author: Afiahayati; Wah Y.B.; Hartati S.; Sari Y.; Trisna I.N.P.; Putri D.U.K.; Musdholifah A.; Wardoyo R.
• Format: Article
• Language: English
• Published: MDPI 2022
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85144725665&doi=10.3390%2fcomputation10120214&partnerID=40&md5=16997a442304404f7179f2ef6f2f16da

Coronavirus disease 2019 (COVID-19) was declared as a global pandemic by the World Health Organization (WHO) on 12 March 2020. Indonesia is reported to have the highest number of cases in Southeast Asia. Accurate prediction of the number of COVID-19 cases in the upcoming few days is required as one of the considerations in making decisions to provide appropriate recommendations in the process of mitigating global pandemic infectious diseases. In this research, a metaheuristics optimization algorithm, the flower pollination algorithm, is used to forecast the cumulative confirmed COVID-19 cases in Indonesia. The flower pollination algorithm is a robust and adaptive method to perform optimization for curve fitting of COVID-19 cases. The performance of the flower pollination algorithm was evaluated and compared with a machine learning method which is popular for forecasting, the recurrent neural network. A comprehensive experiment was carried out to determine the optimal hyperparameters for the flower pollination algorithm and recurrent neural network. There were 24 and 72 combinations of hyperparameters for the flower pollination algorithm and recurrent neural network, respectively. The best hyperparameters were used to develop the COVID-19 forecasting model. Experimental results showed that the flower pollination algorithm performed better than the recurrent neural network in long-term (two weeks) and short-term (one week) forecasting of COVID-19 cases. The mean absolute percentage error (MAPE) for the flower pollination algorithm model (0.38%) was much lower than that of the recurrent neural network model (5.31%) in the last iteration for long-term forecasting. Meanwhile, the MAPE for the flower pollination algorithm model (0.74%) is also lower than the recurrent neural network model (4.8%) in the last iteration for short-term forecasting of the cumulative COVID-19 cases in Indonesia. This research provides state-of-the-art results to help the process of mitigating the global pandemic of COVID-19 in Indonesia. © 2022 by the authors.