Research Article
Unified Quantile Regression Deep Neural Network with Time-Cognition for Probabilistic Residential Load Forecasting
Table 4
AQS comparison of three time-coding methods.
| Residence | Nature coding | One-hot coding | Periodic coding | I_nature (%) | I_one-hot (%) |
| 1 | 0.0604 | 0.0592 | 0.0546 | 9.67 | 7.83 | 2 | 0.1261 | 0.1226 | 0.1206 | 4.40 | 1.67 | 3 | 0.1401 | 0.1449 | 0.1392 | 0.70 | 3.94 | 4 | 0.1823 | 0.1585 | 0.1536 | 15.75 | 3.10 | 5 | 0.0778 | 0.0666 | 0.0651 | 16.25 | 2.15 | 6 | 0.0885 | 0.0951 | 0.0856 | 3.29 | 9.99 | 7 | 0.1413 | 0.1378 | 0.1308 | 7.49 | 5.12 | 8 | 0.2379 | 0.2114 | 0.2101 | 11.66 | 0.60 | 9 | 0.0888 | 0.0895 | 0.0811 | 8.68 | 9.42 | 10 | 0.1091 | 0.1018 | 0.0983 | 9.91 | 3.47 | 11 | 0.0523 | 0.0475 | 0.0461 | 11.89 | 2.91 | 12 | 0.1176 | 0.1116 | 0.1078 | 8.34 | 3.38 | 13 | 0.0433 | 0.0401 | 0.0384 | 11.27 | 4.21 | 14 | 0.0983 | 0.0946 | 0.0901 | 8.26 | 4.67 | 15 | 0.0933 | 0.0896 | 0.0809 | 13.27 | 9.75 | 16 | 0.1188 | 0.1239 | 0.1140 | 4.03 | 8.00 | 17 | 0.0259 | 0.0253 | 0.0226 | 12.89 | 10.63 | 18 | 0.0707 | 0.0630 | 0.0632 | 10.63 | ā0.30 | 19 | 0.1200 | 0.1204 | 0.1091 | 9.09 | 9.34 | 20 | 0.0351 | 0.0349 | 0.0319 | 9.01 | 8.51 | Average | 0.1014 | 0.0969 | 0.0922 | 9.10 | 4.91 |
|
|