Reproduce “AI-BAAM: AI-Driven Bank Statement Analytics as Alternative Data for Malaysian MSME Credit Scoring” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chun Chet Ng, Zhen Hao Chu, Jia Yu Lim, Yin Yin Boon, Wei Zeng Low, Jin Khye Tan
This study introduces AI-BAAM, an AI-driven framework for assessing Malaysian MSME creditworthiness using bank statement analytics, which yielded a blended Logistic Regression model achieving an AUROC of 0.806 for default prediction. The work also established a specialized template matching method for bank statement extraction, demonstrating 100% accuracy and superior efficiency compared to LLM-based approaches.
Key insights
- Bank statement transaction data provides powerful predictive signals for MSME credit default, significantly improving credit scoring accuracy when blended with traditional application information.
- A simpler Logistic Regression model consistently outperformed more complex ensemble methods on the relatively small and class-imbalanced MSME dataset, suggesting its robustness for this specific application.
- A specialized template matching approach for bank statement extraction achieves superior accuracy, lower latency, and zero cost compared to generic LLM-based and end-to-end vision models for structured financial documents.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.