Research on a Rural Credit Risk Assessment Model Based on Machine Learning and Blockchain

View at publisher Summarise with VIRAT
Authors: Zixi Liang
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

To address the challenges of data silos and the inadequate capture of key features by traditional algorithms in rural credit risk assessment, this paper designs a rural credit risk assessment model based on machine learning and blockchain. The model adopts a four-layer progressive architecture: data layer - blockchain layer - machine learning layer - application layer. The blockchain layer, based on the Hyperledger Fabric consortium chain, establishes a data sharing and evidence storage mechanism involving four types of nodes, including financial institutions and governments. Data credibility is ensured through hybrid encryption (AES-256 and SM2) and the PBFT consensus algorithm. The machine learning layer proposes an improved random forest algorithm (F-RFA) that incorporates a feature attention mechanism, optimizing model performance through feature weight calculation and adaptive selection. The experiment used 10,000 rural credit data records from a certain province between 2018 and 2023 (containing 34 features and 99.1% data integrity). The data was divided into a training set (7,000 records) and a test set (3,000 records) with a 7:3 ratio. The performance of F-RFA was compared with that of RF, SVM, and BPNN algorithms, and the effectiveness of the attention mechanism and blockchain sharing was verified. The results showed that F-RFA achieved an accuracy of 92.3%, an improvement of 5.6, 9.8, and 7.5 percentage points over RF, SVM, and BPNN, respectively. Removing the attention mechanism reduced the model's accuracy to 88.5%. Increasing the degree of blockchain data sharing from 0% to 100% increased the model's accuracy by 4.4 percentage points. The study demonstrated that the proposed model can effectively improve the accuracy of rural credit risk assessment and data credibility, providing support for financial institutions' decision-making.

Fields of Study & Subjects

Financial Distress and Bankruptcy Prediction Blockchain Technology Applications and Security Microfinance and Financial Inclusion

Publication details

Journal / source
IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
Pages
700–705
Publication date
28 January 2026
Language
EN