Table 4 Comparison of the forecast performance of each model
Model | RMSE | MAE | MAPE (%) | sMAPE (%) |
|---|---|---|---|---|
6-step ahead prediction between January 2016 to June 2016 | ||||
 A | 0.3244 (−) | 0.2811 (−) | 6.0454 (−) | 5.8097 (−) |
 B | 1.0157 (+ 213.10%) | 0.9339 (+ 232.23%) | 20.0035 (+ 230.89%) | 17.9006 (+ 208.12%) |
 \(\varvec{C}^{\star }\) | 0.2825 (− 12.92%) | 0.2363 (− 15.94%) | 5.0797 (− 15.97%) | 4.9490 (− 14.81%) |
 \(D^{\star }\) | 0.4659 (+ 43.62%) | 0.3206 (+ 14.05%) | 6.8661 (+ 13.58%) | 7.4156 (+ 27.64%) |
12-step ahead prediction between January 2016 and December 2016 | ||||
 A | 0.4425 (−) | 0.3917 (−) | 9.7674 (−) | 9.1462 (−) |
 B | 0.7825 (+ 63.40%) | 0.6508 (+ 66.15%) | 14.5400 (+ 48.86%) | 13.2301 (+ 44.65%) |
 \(C^{\star }\) | 0.4060 (− 8.25%) | 0.3073 (− 21.55%) | 7.8076 (− 20.06%) | 7.5203 (− 17.78%) |
 \(\varvec{D}^{\star }\) | 0.3753 (− 15.19%) | 0.2619 (− 33.14%) | 6.1742 (− 36.79%) | 6.4240 (− 29.76%) |
24-step ahead prediction between January 2016 and December 2017 | ||||
 A | 0.4672 (−) | 0.4177 (−) | 9.9328 (−) | 9.3198 (−) |
 B | 0.7634 (+ 63.40%) | 0.6384 (+ 52.84%) | 14.1518 (+ 42.48%) | 13.0495 (+ 40.02%) |
 \(C^{\star }\) | 0.4108 (− 12.07%) | 0.3295 (− 21.12%) | 7.7436 (− 22.04%) | 7.4895 (− 19.64%) |
 \(\varvec{D}^{\star }\) | 0.4042 (− 13.48%) | 0.3070 (− 26.50%) | 6.9664 (− 29.86%) | 7.2245 (− 22.48%) |