Table 1 Graph neural network-based models show relatively high performance compared to the traditional machine learning methods.

Predicting orders	LR	SVM	NN	GCN	GAT
2nd order prediction
Accuracy	0.6250	0.6250	0.6250	0.6250	0.6250
AUC	0.4000	0.5333	0.4000	0.5333	0.7333
\(F_{1}\) score	0.7692	0.5714	0.7692	0.7692	0.7692
3rd order prediction
Accuracy	0.9038	0.9231	0.9135	0.9327	0.9327
AUC	0.6276	0.4466	0.6224	0.8971	0.7982
\(F_{1}\) score	0.9495	0.9600	0.9548	0.9645	0.9648
4th order prediction
Accuracy	0.9178	0.7808	0.7534	0.8904	0.9315
AUC	0.4859	0.7465	0.7746	0.9648	0.6559
\(F_{1}\) score	0.9571	0.8769	0.8548	0.9403	0.9645

LR logistic regression, SVM support vector machine, NN neural network, GCN graph convolution network, GAT graph attention network, AUC area under the curve.

Quick links

Search