Complexity

Research Article

Predicting and Investigating the Permeability Coefficient of Soil with Aided Single Machine Learning Algorithm

Comparison between ML models in previous studies and the GB model proposed in this study for predicting the permeability coefficient of soil.


Reference	Best machine learning algorithm	Number of inputs	Dataset size	Best performance evaluation for testing part

Pham et al. [33]	RF	6 inputs: water content, void ratio, specific density, liquid limit, plastic limit, and clay content	84	R² = 0.724
				RMSE = 0.840 × 10⁻¹¹ m/s
				MAE = 0.490 × 10⁻¹¹ m/s
Pham et al. [34]	M5P	6 inputs: water content, void ratio, specific density, liquid limit, plastic limit, and clay content	84	R² = 0.766
				RMSE = 0.810 × 10⁻¹¹ m/s
				MAE = 0.450 × 10⁻¹¹ m/s
Bui et al. [35]	TLBO-ANN	6 inputs: water content, void ratio, specific density, liquid limit, plastic limit, and clay content	84	R² = 0.819
				RMSE = 0.294 × 10⁻¹¹ m/s
				MAE = 0.231 × 10⁻¹¹ m/s
Ahmad et al. [36]	GPR-PUK	6 inputs: water content, void ratio, specific density, liquid limit, plastic limit, and clay content	84	R² = 0.951
				RMSE = 0.620 × 10⁻¹¹ m/s
				MAE = 0.370 × 10⁻¹¹ m/s
This study	GB	7 inputs: water content, void ratio, specific density, liquid limit, plastic limit, and clay content + plasticity index	84	R² (mean value of 10000 runs) = 0.804
				R² = 0.971
				RMSE = 0.199 × 10⁻¹¹ m/s
				MAE = 0.161 × 10⁻¹¹ m/s
				MAPE = 0.185