Loan Default Prediction, Filtered Feature Selection, Model Comparison, Adaboost
Authors:
Huang, Yifan
Shao, Yanxi
Tang, Dapeng
Huang, Jie
Chen, Sijia
Journal:
IJIRES
Volume:
10
Number:
3
Pages:
149-159
Month:
May
ISSN:
2349-5219
BibTex:
Abstract:
In recent years, with the booming development of modern technology and information technology, intelligent risk control has become an indispensable part of the healthy development of the financial industry. The core issue in the field of intelligent risk management is to accurately identify potential risks in loans, not to issue loans to borrowers with a high default rate, or to track users who have issued loans in real time, so as to more effectively ensure the interests of lending institutions. Therefore, it is a very important research topic to use the massive data information of lenders and data mining technology to predict the default behavior of loan users. According to the characteristics of unbalanced loan data categories and high feature dimensions, we clean the data and select the features with strong predictive ability by filtered feature selection. Subsequently, based on the comparison of various models, this paper selects the best performing Adaboost model for loan default prediction model construction and conducts model evaluation. The analysis found that all the indicators of Adaboost are high, indicating that the model has better performance and can be used to accurately predict loan defaults. This is used as a basis to provide reference for commercial banks and other lending platforms when offering credit products to borrowers.