notesum.ai

Published at November 22

Financial Risk Assessment via Long-term Payment Behavior Sequence Folding

cs.CY

cs.AI

Released Date: November 22, 2024

Authors: Yiran Qiao¹, Yateng Tang², Xiang Ao¹, Qi Yuan¹, Ziming Liu², Chen Shen², Xuehao Zheng²

Aff.: ¹University of Chinese Academy of Sciences, Beijing 100049, China; ²Tencent Weixin Group, Shenzhen, China

Arxiv: http://arxiv.org/abs/2411.15056v1

Refer to caption

Type	Model	45 days		90 days		180 days
Type	Model	AUC	Recall@10%	AUC	Recall@10%	AUC	Recall@10%
Basic	XGBoost	——		0.6774	0.2556	——
Basic	MLP	——		0.6841	0.2582	——
Sequential	BiLSTM	0.6901	0.2632	0.6914	0.2711	0.7011	0.2794
	Transformer	0.7062	0.2697	0.7139	0.2808	0.7303	0.2931
	Longformer	0.7203	0.2878	0.7357	0.3023	0.7360	0.3002
	Reformer	0.7036	0.2596	0.7091	0.2760	0.7120	0.2672
	Informer	0.7109	0.2813	0.7183	0.2769	0.7230	0.2783
	STAN	0.7105	0.2695	0.7201	0.2783	0.7177	0.2760
	RetNet	0.7216	0.2861	0.7373	0.3016	0.7513	0.3141
	RWKV	0.7171	0.2843	0.7241	0.2811	0.7448	0.3023
	Mamba	0.7286	0.2926	0.7449	0.3109	0.7554	0.3208
Pre-training	BERT4Rec	0.7079	0.2672	0.7215	0.2838	0.7334	0.3046
Pre-training	S3-Rec	0.7131	0.2808	0.7280	0.2917	0.7333	0.3005
Ours	LBSF	0.7240	0.2924	0.7575	0.3259	0.7678	0.3467