Table 7 Results of multi-step ahead EC forecasting for Barratta Creek.

From: Multi-step ahead forecasting of electrical conductivity in rivers by using a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model enhanced by Boruta-XGBoost feature selection algorithm

Horizon

Data

R

RMSE

MAPE

E

Tstat

U95

3-steps-ahead

Train

0.8339

65.6887

14.9481

0.6929

2.2169

182.0097

Test

0.7663

72.4667

14.7316

0.5866

0.8251

200.8821

5-steps-ahead

Train

0.7878

72.9874

16.5071

0.6195

2.7618

202.1789

Test

0.7108

79.4455

16.5628

0.5032

1.7305

220.1116

7-steps-ahead

Train

0.7651

76.1118

16.7717

0.5852

1.0503

210.9690

Test

0.6690

84.2257

16.7168

0.4416

2.6831

233.1329

10-steps-ahead

Train

0.7314

81.0510

17.3607

0.5284

5.6692

223.9778

Test

0.6190

88.6057

18.0405

0.3820

1.4281

245.5441

Back to article page