Prediction of COVID-19 Confirmed Cases after Vaccination: Based on Statistical and Deep Learning Models

In this paper, we analyze and predict the number of daily confirmed cases coronavirus (COVID-19) based on two statistical models a deep learning (DL) model; autoregressive integrated moving average (ARIMA), generalized conditional heteroscedasticity (GARCH), stacked long short-term memory neural network (LSTM DNN). We find orders by autocorrelation function partial function, hyperparameters DL model, such as numbers LSTM cells blocks cell, exhaustive search. Ten datasets are used in experiment; nine countries world datasets, from Dec. 31, 2019, to Feb. 22, 2021, provided WHO. investigate effects data size vaccination performance. Numerical results show that performance depends data's dates vaccination. It also shows prediction DNN is better than those models. Based experimental results, percentage improvements up 88.54% (86.63%) 90.15% (87.74%) compared ARIMA GARCH, respectively, mean absolute error (root squared error). While performances GARCH varying according datasets. The obtained may provide criterion for ranges accuracy COVID-19 cases.Doi: 10.28991/SciMedJ-2021-0302-7 Full Text: PDF