Capítulo de livro
An automated literature analysis on data mining applications to credit risk assessment
Sérgio Moro (Moro, S.); Paulo Cortez (Cortez, P.); Paulo Rita (Rita, P.);
Título Livro
Artificial intelligence in financial markets: Cutting edge applications for risk management, portfolio optimization and economics
Ano (publicação definitiva)
2016
Língua
Inglês
País
Estados Unidos da América
Mais Informação
Web of Science®

Esta publicação não está indexada na Web of Science®

Scopus

Esta publicação não está indexada na Scopus

Google Scholar

N.º de citações: 11

(Última verificação: 2024-03-25 14:47)

Ver o registo no Google Scholar

Abstract/Resumo
This chapter presents an automated literature analysis of data mining applications to credit risk assessment, encompassing the period from 2010 to 2014. Google Scholar was used to collect the 100 most relevant articles published in management and information systems conferences and journals containing the keywords ‘data mining’ and ‘credit risk’. This set of articles served as a basis for assessing the main trends of research in data mining applications to credit risk, first by using text mining, then through the Latent Dirichlet allocation Algorithm for grouping the articles into logical topics. Five types of problems in credit risk were assessed: credit scoring, bankruptcy, credit fraud, credit cards and regulatory issues. From these, credit scoring receives most attention, while bankruptcy and credit fraud were the topic of a significant number of articles. The most interesting finding is that the most advanced data mining techniques such as support vector machines and ensembles are being applied to credit risk problems more for tuning these techniques than to benefit credit risk assessment. This represents an interesting research gap to be addressed. The trends identified prove the value of the automated procedure undertaken, which is novel in credit risk applications. Credit scoring was confirmed as the dominant subject regarding data mining applications. Several studies focused on tuning data mining techniques rather than on showing the benefits achieved by applying such techniques. More focus should be given to the value of data mining to risk assessment. Also, findings suggest that regulatory issues are demanding research in data quality, in alignment with banking regulation leveraged by the global crisis.
Agradecimentos/Acknowledgements
--
Palavras-chave
Support vector machine,Credit card,Credit risk,Text mining,Latent Dirichlet allocation