Table 2 Candidate solutions’ coding.
From: Short-term streamflow modeling using data-intelligence evolutionary machine learning models
Estimator | IP | Description | Settings/range |
|---|---|---|---|
EN | \(\theta _{1}\) | Penalty term, \(\alpha \) | [\(10^{-6}, 2\)] |
\(\theta _{2}\) | \(L_1\)-ratio parameter, \(\rho \) | [0,1] | |
ELM | \(\theta _{1}\) | No. neurons in the hidden layer, L | [1, 500] |
\(\theta _{2}\) | Regularization parameter C | [0.0001, 10000] | |
\(\theta _{3}\) | Activation function G | 1: Identity; 2: Sigmoid; 3: Hyperbolic Tangent; 4: Gaussian; 5: Swish; 6: ReLU; | |
SVR | \(\theta _{1}\) | Loss parameter, \(\varepsilon \) | [10\(^{-5}\), 100] |
\(\theta _{2}\) | Regularization parameter, C | [1, 10000] | |
\(\theta _{3}\) | Bandwidth parameter, \(\gamma \) | [0.001, 10] | |
MARS | \(\theta _{1}\) | Degree of piecewise polynomials, q | [0,3] |
\(\theta _{2}\) | Penalty factor, \(\gamma \) | [1, 9] | |
\(\theta _{3}\) | Maximum number of terms, M | [1, 500] | |
XGB | \(\theta _{1}\) | Learning rate, \(\eta \) | [10\(^{-6}\), 1] |
\(\theta _{2}\) | No. weak estimators, \(M_{est}\) | [10, 500] | |
\(\theta _{3}\) | Maximum depth, \(m_{depth}\) | [1, 20] | |
\(\theta _{4}\) | Regularization parameter, \(\lambda _{reg}\) | [0, 100] |
