Table 1 - Statistical Performance of the Four Forecasting Methods
Table 1 - Statistical Performance of the Four Forecasting Methods
| Model method | Naive model | Long-term memory model | ARIMA model | Nonlinear model |
|---|---|---|---|---|
| RMSE | 597.8 | 904.5 | 637.7 | 471.5 |
| RMSE | (154.7) | (158.8) | (184.7) | (107.9) |
| MAPE | 9.16 | 13.92 | 9.41 | 7.59 |
| MAPE | (1.96) | (2.33) | (2.08) | (2.11) |
| Rank correlation of original | 0.768 | 0.59 | 0.757 | 0.809 |
| Rank correlation of original | (0.087) | (0.093) | (0.090) | (0.071) |
| Significant number of positive errors, 5% | 4 | 11 | 0 | 56 |
| Significant number of positive errors, 10% | 11 | 22 | 4 | 70 |
| Wilcoxon location on original, 5% | 0 | 0 | 0 | 1 |
| Wilcoxon location on original, 10% | 0 | 0 | 0 | 7 |
| Wilcoxon variance on original1, 5% | 0 | 2 | 0 | 14 |
| Wilcoxon variance on original1, 10% | 0 | 4 | 0 | 34 |
| Rank correlation of difference | 0.064 | 0.113 | 0.082 | 0.098 |
| Rank correlation of difference | (0.138) | (0.130) | (0.121) | (0.119) |
| p(Bin)<5% (much better than chance), p(χ2)>10% | 19 | 54 | 21 | 17 |
| p(Bin)<5% (much better than chance), p(χ2)>5% | 19 | 51 | 19 | 15 |
| p(Bin)<10% (better than chance), p(χ2)>10% | 28 | 76 | 30 | 30 |
| p(Bin)<10% (better than chance), p(χ2)>5% | 28 | 72 | 25 | 28 |
1 The number of days on which this test is valid is 184 minus the number of days on which there was a significant difference in the locations of the original and forecast series, e.g., for nonlinear at the 10% level this is 184-7=177 days.