Table 6 The best features of the assessed artificial intelligence models.
Model | Best features | Note |
|---|---|---|
MLPNN | 19 hidden neurons, tangent hyperbolic, and logarithm sigmoid activation functions in the hidden and output layers, respectively | Train function: Levenberg–Marquardt |
CFFNN | 19 hidden neurons, tangent hyperbolic, and logarithm sigmoid activation functions in the hidden and output layers, respectively | Train function: Levenberg–Marquardt |
RBFNN | 19 hidden neurons Gaussian and linear activation functions in the hidden and output layers, respectively Spread value = 0.5263 | – |
GRNN | 550 hidden neurons Gaussian and linear activation functions in the hidden and output layers, respectively Spread = 0.0394 | – |
ANFIS | Radii = 0.435 | Hybrid algorithm (combination of backpropagation scenario and least square method) |
LSSVR | Gaussian kernel | Combination of simulated annealing and simplex optimization methods |
