Table 1
Results from the algorithm used to determine the best ANN architecture

Error from training12.07%
Error from validation30.00%
Error from test25.00%
Error from all data18.37%
Mean Square Error (MSE)0.023206
Training Data60%
Validation Data20%
Test Data20%
Number of neurons in the first layer18
Number of neurons in the second layer13
Function in the first layerpurelin
Function in the second layerlogsig
Training functiontrainscg