Financial Evaluation Model and Algorithm Based on Data Mining

Abstract

With the development of information technology, the traditional financial industry has also entered a period of rapid development. The business scope of financial institutions has expanded dramatically with the technological updates, and the service level and user experience have become higher and higher. However, new credit risk issues inevitably emerge within various areas of the financial market, such as the lending business. The lending business, one of the core businesses of the financial industry, generates huge profits for financial institutions, but is very dependent on the level of risk control. In order to minimize the risk, financial institutions want to use the emerging internet technology to analyze massive data, mine effective information and refine risk indices. Therefore, how to use emerging technologies such as big data and data mining to assess loan defaults is gradually becoming a hot issue for financial institutions and an important research direction. In this paper, 150,000 data records of loan customers are obtained from Kaggle credit score dataset, and data pre-processing is performed by statistical methods to clean the unreasonable data in the dataset, such as duplicate, missing and abnormal values. Using logistic regression algorithm, an interpretable credit evaluation model was built on the user's credit records to predict the default likelihood of the user in the coming years. The final quantitative scoring of loan users' default likelihood helps financial institutions control their risks.