Table 3. Accuracy measures of the models based on a split in training set (January 2015 to February 2019) and test sets (March 2019 to December 2019) for the Brazilian data.
| ME | RMSE | MAE | MPE | MAPE | ACF1 | Theil’s U | |
|---|---|---|---|---|---|---|---|
| Training set (ARIMA) | 568.54 | 2525.20 | 1532.22 | 0.50% | 1.45% | -0.002 | 0.49 |
| Test set (ARIMA) | 2094.77 | 2465.82 | 2302.43 | 1.86% | 2.05% | 0.34 | |
| Training set (ETS) | 81.32 | 2172.97 | 1610.00 | 0.03% | 1.55% | 0.02 | 0.58 |
| Test set (ETS) | 2148.03 | 3005.03 | 2667.15 | 1.89% | 2.37% | 0.53 | |
| Training set (ALNB) | 806.72 | 4111.25 | 2883.05 | 0.60% | 2.78% | 0.12 | 0.67 |
| Test set (ALNB) | 3307.83 | 3567.81 | 3307.83 | 2.93% | 2.93% | -0.29 |