COVID-19 in Iran: Forecasting Pandemic Using Deep Learning
Table 4
Comparison of performance in test stage on nine selected countries.
Model
BI
DI
Lag
MAPE
RMSE
NRMSE
RF
2.069
0.980306
2823
0.001057
2.557
0.979987
3012
0.001128
2.580
0.979985
4620
0.00173
MLP
0.337
0.999989
1059
0.000397
1.686
0.980184
4634
0.001735
0.786
0.989893
3046
0.001141
LSTM-R
2.030
0.980005
4963
0.0018658
1.941
0.980102
4702
0.001761
1.536
0.986131
4369
0.001636
LSTM-E
0.547
0.989981
1784
0.000668
0.853
0.989634
2921
0.001093
0.624
0.989972
1991
0.000745
M-LSTM
0.302
0.999991
1000
0.000374
0.132
0.999997
621
0.000232
0.073
0.999999
210
0.0000786
represents only recovered cases as input and shows the confirmed cases along with death and recovered cases. BI: basic information; DI: detailed information. The performance is evaluated on recovered cases as output.
