DTD-VAE: Disentangled Temporal Dependencies VAE for Credit Risk Prediction
Abstract
Evaluating customer creditworthiness is crucial for retail banking operations, as it impacts marketing strategies, customer relationship management, and credit risk control. Traditional methods often struggle to capture complex temporal dependencies and extract pertinent information from customer data, crucial for accurate risk assessment. Specifically, they fail to differentiate between temporal patterns indicative of credit risk and those reflecting general customer behavior or preferences, leading to suboptimal risk predictions. In this study, we introduce the Disentangled Temporal Dependencies Variational Autoencoder (DTD-VAE), an advancement over conventional VAE, designed to disentangle temporal dependencies and distinguish credit risk-related features from past customer preferences. The feature inference module of the DTD-VAE incorporates an autoregressive temporal dependency learning mechanism that adeptly captures the temporal dependencies among latent variables, enriching the model's comprehension of the inherent data structure. Furthermore, the feature generative module utilizes an element-wise gating mechanism that assigns independent weights to each dimension of the expert models, enabling a finer-grained disentanglement of latent variables, particularly those relevant to credit risk prediction. Extensive experiments on six real-world datasets demonstrate that the proposed framework consistently outperforms existing methods, achieving performance gains of 3.2%-4.86% in ROC-AUC and 6.41%-9.71% in Accuracy Ratio.
Related
- Persistence, Exposure, and Credit Routing in Linear Temporal Predictive Coding
- Research on expressway traffic accident risk prediction method considering Spatio-Temporal effects
- Discrete-Time Survival Modelling with Dynamic Industrial Graphs for Mortgage Credit Risk
- Enhancing credit risk scorecards with transactional-based behavioural features
- BCFL-DAI-PUB: Blockchain-integrated federated learning with dynamic adaptive incentives for privacy-utility balanced credit risk modelling
BibTeX
@misc{guo2026dtdvae,
title = {{DTD-VAE: Disentangled Temporal Dependencies VAE for Credit Risk Prediction}},
author = {Xiaobo Guo and Lu-an Dong and Yanbo Wang and Peng Zhang and Cai Zhi and Youru Li},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.26473},
eprint = {2608.26473},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.26473},
}
RIS
TY - UNPB
TI - DTD-VAE: Disentangled Temporal Dependencies VAE for Credit Risk Prediction
AU - Xiaobo Guo
AU - Lu-an Dong
AU - Yanbo Wang
AU - Peng Zhang
AU - Cai Zhi
AU - Youru Li
PY - 2026
DA - 2026/08/31
PB - arXiv
DO - 10.48550/arxiv.2608.26473
UR - https://arxiv.org/abs/2608.26473
AB - Evaluating customer creditworthiness is crucial for retail banking operations, as it impacts marketing strategies, customer relationship management, and credit risk control. Traditional methods often struggle to capture complex temporal dependencies and extract pertinent information from customer data, crucial for accurate risk assessment. Specifically, they fail to differentiate between temporal patterns indicative of credit risk and those reflecting general customer behavior or preferences, leading to suboptimal risk predictions. In this study, we introduce the Disentangled Temporal Dependencies Variational Autoencoder (DTD-VAE), an advancement over conventional VAE, designed to disentangle temporal dependencies and distinguish credit risk-related features from past customer preferences. The feature inference module of the DTD-VAE incorporates an autoregressive temporal dependency learning mechanism that adeptly captures the temporal dependencies among latent variables, enriching the model's comprehension of the inherent data structure. Furthermore, the feature generative module utilizes an element-wise gating mechanism that assigns independent weights to each dimension of the expert models, enabling a finer-grained disentanglement of latent variables, particularly those relevant to credit risk prediction. Extensive experiments on six real-world datasets demonstrate that the proposed framework consistently outperforms existing methods, achieving performance gains of 3.2%-4.86% in ROC-AUC and 6.41%-9.71% in Accuracy Ratio.
ER -