Data Science in FinTech: credit risk prediction using Deep Learning

Abstract

Data science has been a hot field in recent years, and many companies with large amounts of data are interested in applying data science techniques to make a profit from their data. The possibilities with data science are, therefore, very large, and Fintech, which is an emerged industry, is trying to make use of this hot field to change the traditional way of providing financial services. FinTech companies can use data science to make predictions (predictive modeling) and extract knowledge from many structured and unstructured data. There are many ways to extract valuable information from large amounts of data that FinTech companies own. This thesis focuses on one of the axes where data science and FinTech meet, which is credit risk modeling. This is done by implementing a typical data science project, where deep learning techniques are applied to predict the ability of borrowers to repay their loans. This case of study follows the important steps of a data science project cycle to solve the problem presented by the financial institution Home Credit through the Kaggle platform. Each step is discussed and elaborated in this case study. In addition, the goal is to predict whether future Home Credit customers will be able to repay their requested loan with or without payment difficulties. Our deep neural network model was able to get a good score, which will help Home Credit to make their predictions. The thing that will encourage others to do more research and let them apply other models and try other data science techniques to solve a lot of problems related to the financial field.