An empirical comparison of classification algorithms for mortgage default prediction: evidence from a distressed mortgage market

Fitzpatrick, Trevor; Mues, Christophe

doi:10.1016/j.ejor.2015.09.014

Cited by 77 publications

(30 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A lot of research effort has been committed to evaluating classification algorithms in credit scoring, ranging from traditional statistical methods, such as logistic regression [1], to non-parametric algorithms, such as neural networks [9]. In the recent years there has been an increased interest for using hybrid and ensemble classifiers in credit risk, such as boosted regression trees, random forests, deep learning methods and other [10], [11], [12], [13] [14]. A number of benchmark studies have been performed, comparing classification accuracy of different classification algorithms [15], [8].…”

Section: Background Workmentioning

confidence: 99%

An insight into the effects of class imbalance and sampling on classification accuracy in credit risk assessment

Andric¹,

Kalpić²,

Bohaček³

2019

COMSIS J

View full text Add to dashboard Cite

In this paper we investigate the role of sample size and class distribution in credit risk assessments, focusing on real life imbalanced data sets. Choosing the optimal sample is of utmost importance for the quality of predictive models and has become an increasingly important topic with the recent advances in automating lending decision processes and the ever growing richness in data collected by financial institutions. To address the observed research gap, a large-scale experimental evaluation of real-life data sets of different characteristics was performed, using several classification algorithms and performance measures. Results indicate that various factors play a role in determining the optimal class distribution, namely the performance measure, classification algorithm and data set characteristics. The study also provides valuable insight on how to design the training sample to maximize prediction performance and the suitability of using different classification algorithms by assessing their sensitivity to class imbalance and sample size.

show abstract

Section: Background Workmentioning

confidence: 99%

An insight into the effects of class imbalance and sampling on classification accuracy in credit risk assessment

Andric¹,

Kalpić²,

Bohaček³

2019

COMSIS J

View full text Add to dashboard Cite

show abstract

“…Fitzpatrick and Mues () evaluate the performance of several modeling approaches for determining future mortgage default status. Boosted regression trees, random forests, penalized linear and semi‐parametric logistic regression models are applied to four portfolios covering 300,000 Irish owner–occupier mortgages.…”

Section: Previous Studiesmentioning

confidence: 99%

Developing a Risk Group Predictive Model for Korean Students Falling into Bad Debt

Han¹,

Choi²,

Kim³

et al. 2018

Asian Economic Journal

View full text Add to dashboard Cite

Using direct loan data for 2012 to 2014 from the Korea Student Aid Foundation, we develop a risk group predictive model for borrowers defaulting on their loans. We used a logistic regression model and the Cox proportional hazards model to develop the risk predictive model. We verified the validity of the models using a receiver operating characteristic curve and a validation dataset. The present study shows that area under the receiver operating characteristic curves is similar for the models and that the major influencing factors for defaulting on their loans are household income, whether a national grant was received, age, whether more than two accounts are overdue, field of study and the monthly repayment amount. The risk group predictive model in this study will be the basis for more efficient management of direct student loans.

show abstract

“…Our study is a shift from the previous (public) machine learning studies on consumer loans as our dataset consist of extensive amounts of real data. Other machine learning-based studies conducted to predict defaults in consumer loans include Khandani et al [4], Butaru et al [5], and Fitzpatrick and Mues [6]. Khandani et al [4] used generalized classification and regression trees to construct nonlinear, nonparametric forecasting models of consumer credit risk by combining customer transactions and credit bureau scores.…”

Section: Introduction and Literature Reviewmentioning

confidence: 99%

“…Butaru et al [5] applied logistic regression, decision trees using the C4.5 algorithm, and the random forests methods to combined consumer trade-line, credit-bureau, and macroeconomic variables to predict delinquency. Fitzpatrick and Mues evaluated the performance of logistic regression, semiparametric generalised additive models, boosted regression trees, and random forests for future mortgage default status [6]. These studies aim to give an overview of the objectives, techniques, and difficulties of credit scoring as an application of forecasting.…”

Section: Introduction and Literature Reviewmentioning

confidence: 99%

Consumer loans’ first payment default detection: a predictive model

Koc¹,

Sevgili²

2020

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

A default loan (also called nonperforming loan) occurs when there is a failure to meet bank conditions and repayment cannot be made in accordance with the terms of the loan which has reached its maturity. In this study, we provide a predictive analysis of the consumer behavior concerning a loan's first payment default (FPD) using a real dataset of consumer loans with approximately 600,000 records from a bank. We use logistic regression, naive Bayes, support vector machine, and random forest on oversampled and undersampled data to build eight different models to predict FPD loans. A two-class random forest using undersampling yielded more than 86% on all performance measures: accuracy, precision, recall, and F1-score. The corresponding scores are even as high as 96% for oversampling. However, when tested on the real and balanced dataset, the performance of oversampling deteriorates as generating synthetic data for an extremely imbalanced dataset harms the training procedure of the algorithms. The study also provides an understanding of the reasons for nonperforming loans and helps to manage credit risks more consciously.

show abstract

An empirical comparison of classification algorithms for mortgage default prediction: evidence from a distressed mortgage market

Cited by 77 publications

References 52 publications

An insight into the effects of class imbalance and sampling on classification accuracy in credit risk assessment

An insight into the effects of class imbalance and sampling on classification accuracy in credit risk assessment

Developing a Risk Group Predictive Model for Korean Students Falling into Bad Debt

Consumer loans’ first payment default detection: a predictive model

Contact Info

Product

Resources

About