Combined machine learning approaches to predict the thermal conductivity of liquid mixtures | Science and Technology for Energy Transition (STET)

Open Access

Table 6

R ², RMSE (in W m⁻¹ K⁻¹), and MAE (in W m⁻¹ K⁻¹) values calculated for the different mixing rules powered with the three QSPR models, when applied to the DB dataset.

QSPR model	Mixing Rule	R²	RMSE	MAE	R²	RMSE	MAE
		Exp + Pred			Full Pred
Moreno’s SVR [3]	Equation (11)	0.935	0.0054	0.0039	0.903	0.0066	0.0051
	Equation (10)	0.933	0.0055	0.0039	0.903	0.0066	0.0051
	Equation (1) with α = 0.72	0.901	0.0067	0.0046	0.870	0.0077	0.0058
	Equation (1) with α = 0.5	0.915	0.0062	0.0044	0.884	0.0072	0.0055
	Equation (2)	0.922	0.0060	0.0041	0.894	0.0069	0.0053
SVR	Equation (12)	0.941	0.0052	0.0032	0.905	0.0066	0.0047
	Equation (10)	0.934	0.0055	0.0035	0.901	0.0067	0.0049
	Equation (1) with α = 0.72	0.925	0.0058	0.0036	0.901	0.0067	0.0048
	Equation (1) with α = 0.5	0.928	0.0057	0.0035	0.902	0.0067	0.0048
	Equation (2)	0.929	0.0057	0.0035	0.898	0.0068	0.0049
XGBoost	Equation (13)	0.971	0.0037	0.0026	0.945	0.0050	0.0039
	Equation (10)	0.966	0.0039	0.0027	0.942	0.0051	0.0039
	Equation (1) with α = 0.72	0.944	0.0050	0.0032	0.924	0.0059	0.0043
	Equation (1) with α = 0.5	0.954	0.0046	0.0030	0.933	0.0055	0.0041
	Equation (2)	0.963	0.0041	0.0028	0.939	0.0052	0.0039

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.