Table 7 Optimized Hyperparameter Tuning for ML models.

From: Prediction of rapid chloride permeability using silica fume, fly ash, GGBS and micro fibers based geopolymer concrete

Model

Key Hyperparameters

Optimized values

AdaBoost

n_estimators, learning rate

n_estimators = 150, learning _rate = 0.8

AVOA

Population size, iterations, and exploitation coefficient

Population = 30, Iterations = 100, α = 0.7

CatBoost

depth, learning_rate, iterations, l2_leaf_reg

depth = 8, learning rate = 0.05, iterations = 800, l2_leaf_reg = 3.0

LightGBM

num_leaves, learning_rate, n_estimators, feature_fraction

num_leaves = 40, learning_rate = 0.07, n_estimators = 500, feature_fraction = 0.8