Table 5.

A summary of different computation methods used for SARS-CoV-2 forecasting.

Studied location	Forecasting Method	Input	Validation/Error method	Results	Reference
Brazil	→Support vector regression (SVR)	→Dataset of confirmed cases of COVID-19	→Cross validation	→Error for SVR →1 day: 0.87-3.51% →2 days: 1.02-5.63% →3 days: 0.95-6.90%	(Ribeiro et al., 2020b)
Canada	→Deep learning long short-term memory (LSTM)	→Number of confirmed cases →Number of fatalities and recovered patient	→Cross validation	→Accuracy: →Short term: 93.45% →Long term: 92.67% →Outbreak end was estimated to be on June 2020	(Chimmula and Zhang, 2020)
China	→Adaptive neuro-fuzzy inference system using enhanced flower pollination method	→World Health Organization (WHO) official data of the outbreak of the COVID-19	→Holdout	→High performance in predicting confirmed cases →R2=0.97	(Al-Qaness et al., 2020)
India USA UK	→Long short-term memory (LSTM) →Memory optimized by Grey Wolf →Optimizer deep learning approach	→Google trends →European Centre for Disease prevention and Control (ECDC) data	→Akaike information criterion	→Reduction in Mean Absolute Percentage Error (MAPE) values for forecasting results to the extent of about 98.00%	(Prasanth et al., 2021)
Italy	→Auto-regressive integrated moving average (ARIMA) Forecasting package	→COVID-19 infected patient data from Italian ministry of health	→Mean absolute prediction parameter	→93.75% of accuracy for registered case models →84.40% of accuracy for recovered case models	(Chintalapudi et al., 2020)
Italy Spain France China Australia USA	→Variational autoencoder (VAE) deep learning	→Daily confirmed and recovered cases collected from six countries	→Holdout	→Error per country →Italy: 5.90% →Spain: 2.19% →France: 1.88% →China: 0.13% →Australia: 0.24% →USA: 2.04%	(Zeroual et al., 2020)
Mexico	→Decision tree algorithim	→Epidemiology dataset by Secretariat of Health in Mexico	→Cross validation	→Accuracy: 94.99%	(Muhammad et al., 2021)
Ukraine	→Polynomial regression	→Daily incidence of coronavirus infection →Population size →Viral propagation speed	→Holdout	→Accuracy: 97.60%	(Chumachenko et al., 2020)
24 Countries and 24 States	→Artificial neural network (ANN)	→Dataset provided by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University	→Handout	→Average accuracy of 87.70%	(Wieczorek et al., 2020b)
12 countries	→Support vector regression (SVR)	→Dataset provided by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University	→Cross validation	→Ability to capture nonlinear patterns from the data →Gaussian Kernel provided best in-sample performance and also provided worst out-of-sample prediction	(Peng and Nagata, 2020)