Clues from networks: quantifying relational risk for credit risk evaluation of SMEs

Financial Innovation

Table 29 Discrimination performance of different feature sets under 90% confidence intervals

Model	Metric	BF + CNAP	BF + CNLD	BF + DNAP	BF + DNLD	BF + SNAP	BF + SNLD
LR	AUC	0.723 ± 0.012	0.729 ± 0.012	0.730 ± 0.011	0.734 ± 0.012	0.725 ± 0.012	0.731 ± 0.011
	KS	0.417 ± 0.018	0.429 ± 0.019	0.427 ± 0.017	0.442 ± 0.018	0.417 ± 0.018	0.425 ± 0.019
	H	0.268 ± 0.017	0.282 ± 0.017	0.269 ± 0.015	0.288 ± 0.017	0.269 ± 0.017	0.278 ± 0.017
RF	AUC	0.736 ± 0.011	0.744 ± 0.011	0.743 ± 0.011	0.747 ± 0.010	0.737 ± 0.011	0.741 ± 0.011
	KS	0.442 ± 0.018	0.457 ± 0.018	0.456 ± 0.018	0.455 ± 0.017	0.444 ± 0.019	0.454 ± 0.019
	H	0.283 ± 0.016	0.296 ± 0.016	0.291 ± 0.015	0.295 ± 0.015	0.282 ± 0.017	0.296 ± 0.017
XGB	AUC	0.713 ± 0.011	0.715 ± 0.010	0.721 ± 0.010	0.732 ± 0.011	0.716 ± 0.011	0.725 ± 0.010
	KS	0.415 ± 0.016	0.410 ± 0.017	0.423 ± 0.016	0.439 ± 0.017	0.415 ± 0.017	0.436 ± 0.017
	H	0.239 ± 0.014	0.250 ± 0.015	0.244 ± 0.014	0.276 ± 0.017	0.244 ± 0.015	0.251 ± 0.015

Each metric are measured in mean of 100 estimates obtained by repeating the outer ten-fold cross-validation procedure ten times (mean ± standard deviation from a 90% confidence interval)