Research Article

A Time-Series Water Level Forecasting Model Based on Imputation and Variable Selection Method

Table 7

The results of compare forecasting models under 10-folds cross-validation after variable selection.

MethodsIndexRBF NetworkKstarRandom ForestIBKRandom Tree

After variable selectionDelete the rows with missing dataCC0.1030.6650.7370.2330.529
MAE0.1810.1150.1080.1930.143
RMSE0.2260.1710.1540.2820.223
RAE0.9840.6270.5891.0470.774
RRSE0.9940.7490.6771.2380.977
Series meanCC0.0810.6880.7510.2950.547
MAE0.1690.1030.0980.1700.131
RMSE0.2170.1580.1440.2600.209
RAE0.9880.6000.5710.9900.767
RRSE0.9960.7270.6611.1930.960
LinearCC0.0810.6920.7500.2860.551
MAE0.1710.1020.0980.1690.128
RMSE0.2180.1580.1450.2610.207
RAE0.9880.5900.5660.9810.740
RRSE0.9960.7230.6621.1960.948
Median of nearby pointsCC0.0830.6920.7520.3050.555
MAE0.1710.1020.0970.1690.126
RMSE0.2180.1580.1440.2590.208
RAE0.9870.5930.5630.9800.732
RRSE0.9960.7220.6601.1860.951
Mean of nearby pointsCC0.0820.6940.7530.2760.537
MAE0.1710.1020.0970.1710.129
RMSE0.2180.1570.1440.2630.210
RAE0.9880.5930.5640.9930.747
RRSE0.9960.7210.6591.2040.960
RegressionCC0.0810.6900.7530.2950.572
MAE0.1690.1020.0980.1690.126
RMSE0.2170.1580.1440.2590.204
RAE0.9880.5950.5710.9890.735
RRSE0.9960.7250.6591.1930.938

denotes after variable selection with enhancing performance; denotes the best performance among 5 models after variable selection.