Research of personal loans based on random forests and support vector machines
Nan, Zhu Meng
Ying, Huang Su
- 经济学院－已发表论文 
Since the U.S. sub-prime mortgage crisis broke in 2008, personal loans risk analysis has attracted lots of research interests, particularly in china. China's real estate market bubble began to appear, the implied default risk of personal housing loan which accounted for the majority proportion in china is gradually exposed. Recently, Artificial Intelligence (AI) methods has been proven to be better performance in solving a variety of risk problems including personal loans risk prediction than traditional statistical methods. In this paper, we introduces a relatively new machine learning technique, Random Forest (RF) and Support Vector Machines (SVM), to the problem in attempt to provide a model achieving better performance in prediction accuracy. We applied Random Forest(RF) to select input financial variables and obtained relative importance of the input financial variables, and based on the input variables, we construct SVM model. Experiment results on personal loans datasets of china and comparison with the performance between the model and other three individual classification models on predictive accuracy of default rate are presented. The experimental evaluation shows that the model based on RF and SVM could improve prediction accuracy.