Table 3.

Comparison of general models.

Method	AUC	Accuracy	Recall	Precision	F1 score
TabNet (20)	0.7891	0.7755	0.7727	0.7391	0.7559
AutoML (21)	0.7453	0.7368	0.7143	0.7895	0.7519
DeepFM (22)	0.6941	0.6818	0.7273	0.6667	0.6970
XGBoost (23)	0.7131	0.7097	0.6429	0.6923	0.6676
Our Model	0.8412	0.8191	0.8597	0.8389	0.8464

For all the comparison methods please refer to the opensource implementations at TabNet: https://github.com/dreamquark-ai/tabnet, AutoML: https://github.com/google/automl, DeepFM: https://github.com/ChenglongChen/tensorflow-DeepFM, XGBoost: https://github.com/dmlc/xgboost.

Bold values indicate the best performed method.