Interpretability of Neural Networks: A Credit Card Default Model Example

Abstract

Neural networks have risen in popularity for a number of applications, also in quantitative finance. However, the low interpretability of their ‘black box’ representation has always been a common criticism. Previous literature has attempted to provide a better understanding and visualisation of neural networks, focusing primarily on image classification. This paper shows the feasibility of applying the same methods to an example deep neural network model, concerned with the estimation of credit risk for a portfolio of credit cards. Results show that the analysis of relevance, sensitivity and neural activities can increase the interpretability of a neural network in a financial modelling context.

Keywords: Machine Learning, Credit Risk, Neural Networks, Credit Cards, Interpretability, Explainability