Research Article
Short-Term Speed Prediction Using Remote Microwave Sensor Data: Machine Learning versus Statistical Model
Table 1
Prediction accuracy of models for different forecasting steps ahead in station A.
| MAE | Number of forecasting steps ahead | 1 | 3 | 5 | 10 |
| BPNN | 2.5027 | 3.0243 | 3.4576 | 4.3553 | NARXNN | 2.5041 | 3.0232 | 3.4482 | 4.3357 | SVM-RBF | 2.6537 | 3.1135 | 3.4858 | 4.2625 | SVM-LIN | 2.7492 | 3.1160 | 3.4714 | 4.2604 | MLR | 3.0820 | 3.4795 | 3.8774 | 4.6338 | ARIMA | 2.6777 | 3.1501 | 3.5291 | 4.3233 | VAR | 2.6835 | 3.2931 | 3.7145 | 4.6030 | ST | 2.9398 | 3.1414 | 3.4932 | 4.2355 |
| MAPE (%) | Number of forecasting steps ahead | 1 | 3 | 5 | 10 |
| BPNN | 5.2831 | 6.4927 | 7.5472 | 9.5735 | NARXNN | 5.2923 | 6.4995 | 7.4855 | 9.6267 | SVM-RBF | 5.5839 | 6.6952 | 7.5901 | 9.4775 | SVM-LIN | 5.2878 | 6.5408 | 7.6922 | 9.4937 | MLR | 5.9048 | 7.3508 | 8.1869 | 9.9782 | ARIMA | 5.6290 | 6.7794 | 7.6974 | 9.6376 | VAR | 5.6271 | 6.7202 | 8.2556 | 9.5937 | ST | 6.1660 | 6.7725 | 7.6390 | 9.5080 |
|
|