Table 7.

Prediction results of the regression methods split by category of dataset. The maximum F-score, Spearman's Correlation and root mean square error are reported. Note that all Spearman correlation values are significant ($p<0.01)$.

Method	Category	Max F-Score	Correlation	RMSE
OLS	VGI	0.31	0.340	5.29E-03
	SM	0.43	0.516	4.82E-03
	AUTH	0.49	0.678	4.32E-03
	COMBO	0.51	0.699	4.02E-03
Spatial Lag	VGI	0.36	0.429	5.13E-03
	SM	0.42	0.518	4.82E-03
	AUTH	0.49	0.637	4.34E-03
	COMBO	0.51	0.694	4.05E-03
SVM	VGI	0.28	0.301	5.35E-03
	SM	0.35	0.417	5.17E-03
	AUTH	0.47	0.582	5.21E-03
	COMBO	0.55	0.590	5.17E-03
	COMBO & LL	0.56	0.587	5.17E-03
RDF	VGI	0.31	0.606	4.69E-03
	SM	0.43	0.849	3.20E-03
	UGC	0.46	0.855	3.32E-03
	AUTH	0.57	0.930	2.25E-03
	COMBO	0.62	0.955	1.85E-03
	COMBO & LL	0.74	0.960	1.79E-03