Benchmarking state-of-the-art classification algorithms for credit scoring

By Bart Baesens, Tony Van Gestel, S. Viaene, M. Stepanova, Johan A. K. Suykens, Jan Vanthienen. Journal of the Operational Research Society, 2003

416 citations recorded by the source corpus.

Subfields: Accounting, Artificial Intelligence, Health Information Management

View Paper: https://www.semanticscholar.org/paper/e759fe8bdc24462c9f40b1829a0fbdd84bcc8e03

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Accounting

In Artificial Intelligence

In Health Information Management

Citations in the corpus

Handling class imbalance in customer churn prediction (2009, result): “…has been shown to provide good and robust results in general comparison studies, for both churn prediction (Neslin et al., 2006) and credit scoring (Baesens et al., 2003) and (4) more specifically in database marketing , it has been shown by several authors (Levin & Zahavi, 1998) that logit…”
Kernel Methods in Finance (2008, background): “(Baesens et al., 2003) provide an extensive study on state-of-the-art classification algorithms and their performance within credit scoring.”
Classification methods applied to credit scoring: A systematic review and overall comparison (2016, methodology): “Baesens et al. (2003) performed a comparison involving discriminant analysis, logistic regression, logic programing, support vector machines, neural networks, bayesian networks, decision trees and k-nearest neighbor.”
Building comprehensible customer churn prediction models with advanced rule induction techniques (2011, methodology): “For the SVM a RBF Kernel is chosen as it is shown to achieve a good overall performance (Baesens et al., 2003; Martens et al., 2009).”
The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients (2009, background): “Baesens et al. (2003) found that, based on eight real-life credit scoring data sets, both the LS-SVM and neural network classifiers yield a very good performance, but also simple classifiers such as logistic regression and linear discriminant analysis perform very well for credit scoring.”
Benchmarking Classification Models for Software Defect Prediction: A Proposed Framework and Novel Findings (2008, background): “It is known that a careful tuning of such hyperparameters is essential to obtain a representative assessment of the classifier’s potential (see, e.g., [3], [63]).”
Domain knowledge integration in data mining using decision tables: case studies in churn prediction (2009, methodology): “The motivation is that both techniques yield simple, easy to understand models and furthermore have been shown to perform well in various problem settings, e.g. for credit scoring (Baesens et al , 2003 b ).”
An experimental comparison of classification algorithms for imbalanced credit scoring data sets (2012, methodology): “In addition, some benchmarking studies have been undertaken to empirically compare the performance of these various techniques (e.g., Baesens et al., 2003), but they did not focus specifically on how these techniques compare on heavily imbalanced samples, or to what extent any such comparison is…”
Classification With Ant Colony Optimization (2007, methodology): “This finding is typical for data mining, as no technique performs best on all data sets [5].”
Extended Taxonomy of Rule Extraction Techniques and Assessment of KDRuleEx (2012, methodology): “Rule extraction techniques can represent extracted model in form of If-Then rules, M of N rules, decision tree, decision table, etc. Opaque models like ANN and SVM shows better accuracy performance in comparison to white box or transparent model like decision tree [1][2][3].”
Forecasting and operational research: a review (2008, background): “Crone et al (2006) review publications in business DM and find a strong emphasis on evaluating and tuning multiple classification algorithms on a single data set, Baesens et al (2003) being an exception.”
Integrating induction and deduction for finding evidence of discrimination (2009, background): “For instance, data mining and machine learning classification models are constructed on the basis of historical data exactly with the purpose of learning the distinctive elements of different classes, such as good/bad debtor in credit/insurance scoring systems [3, 10, 27] or good/bad worker in perso”