Analysis of Travel Mode Choice in Seoul Using an Interpretable Machine Learning Approach

Table 3

Comparison of prediction performances of the ML models.

Travel modes	Specificity			Sensitivity			Balanced accuracy
	ANN	RF	XGB	ANN	RF	XGB	ANN	RF	XGB
Car	0.806	0.881	0.920	0.752	0.713	0.670	0.779	0.797	0.795
Bike	0.985	0.990	0.993	0.116	0.338	0.291	0.550	0.664	0.642
Transit	0.879	0.887	0.883	0.515	0.639	0.744	0.697	0.763	0.813
Walking	0.819	0.834	0.850	0.739	0.826	0.856	0.779	0.830	0.853
All	0.882	0.909	0.923	0.647	0.727	0.768	0.764	0.818	0.845

Note. ANN = artificial neural network; RF = random forest; XGB = extreme gradient boosting.