Predicting mortgage default using convolutional neural networks

By Håvard Kvamme, Nikolai Sellereite, Kjersti Aas, Steffen Sjursen. 2018

155 citations recorded by the source corpus.

Abstract We predict mortgage default by applying convolutional neural networks to consumer transaction data. For each consumer we have the balances of the checking account, savings account, and the credit card, in addition to the daily number of transactions on the checking account, and amount transferred into the checking account. With no other information about each consumer we are able to achieve a ROC AUC of 0.918 for the networks, and 0.926 for the networks in combination with a random forests classifier.

Subfields: Accounting, Artificial Intelligence, Economics and Econometrics

Citations in the corpus

• Deep generative models for reject inference in credit scoring (2019, unknown): “For an application of CNNs in credit scoring the reader is referred to [47].”
• Infusing domain knowledge in AI-based "black box" models for better explainability with application in bankruptcy prediction (2019, unknown): “pers [16, 17, 20, 23, 24, 26, 28, 31–36, 39, 41–44, 48–51, 55, 56, 59, 62– 67, 71, 73] related to mortgage default/bankruptcy prediction and collected all explanatory variables (i.”
• Explaining individual predictions when features are dependent: More accurate approximations to Shapley values (2019, unknown): “The data set consists of 28 features extracted from 6 transaction time series. It has previously been used for predicting mortgage default, relating probability of default to transaction information (Kvamme et al., 2018). The transaction information consists of the daily balances on consumers’ credi”