Applied Intelligence

, Volume 49, Issue 2, pp 555–568 | Cite as

Dynamic weighted ensemble classification for credit scoring using Markov Chain

  • Xiaodong Feng
  • Zhi XiaoEmail author
  • Bo Zhong
  • Yuanxiang Dong
  • Jing Qiu


As the ensemble methods achieve significantly better performances than individual models do, they have been widely applied to credit scoring. However, most of them employ a static combiner to combine base classifiers, which do not consider the base classifiers’ characters and their dynamic classification ability. Though some dynamic ensemble methods are proposed, they need to produce a large number of base classifiers or employ a fixed combiner, which limit the generality of the ensemble methods. In this paper, we propose a new dynamic weighted ensemble method for credit scoring. Markov Chain is employed to model the change of each classifier’s classification ability and build a dynamic weighted trainable combiner, which dynamically assign weights to the base classifiers for each sample in the testing set. Through eight credit data sets from the real world, the experimental study demonstrates the ability and efficiency of the dynamic weighted ensemble method to improve prediction performance against the benchmark models, including some well-known individual classifiers and dynamic ensemble methods. Moreover, the proposed method can effectively decrease the misclassification cost, which can reduce risks for the financial institutions.


Credit scoring Dynamic weighted ensemble Markov chain Machine learning 



We thank the editor and the referees for their constructive remarks that helped to improve the clarity and the completeness of this paper. The work was supported by the National Natural Science Foundation of China [grant numbers 71671019, 71701116]; and MOE (Ministry of Education in China) Project of Humanities and Social Sciences [grant number 15YJC630016].


  1. 1.
    Lin WY, Hu YH, Tsai CF (2012) Machine learning in financial crisis prediction: a survey. IEEE T Syst Man Cy C 42(4):421–436. Google Scholar
  2. 2.
    Bahrammirzaee A, Ghatari AR, Ahmadi P, Madani K (2011) Hybrid credit ranking intelligent system using expert system and artificial neural networks. Appl Intell 34(1):28–46. Google Scholar
  3. 3.
    BCBS (2011) Basel III: a global regulatory framework for more resilient banks and banking systems. Bank for International Settlements, BaselGoogle Scholar
  4. 4.
    Lessmann S, Baesens B, Seow HV, Thomas LC (2015) Benchmarking state-of-the-art classification algorithms for credit scoring: an update of research. Eur J Oper Res 247(1):124–136zbMATHGoogle Scholar
  5. 5.
    Avery RB, Calem PS, Canner GB (2004) Consumer credit scoring: do situational circumstances matter? J Banking Finance 28(4):835–856. Google Scholar
  6. 6.
    Zhou ZH (2008) Knowledge acquisition via ensemble learning. In: 2008 international forum on knowledge technology, pp 361–362Google Scholar
  7. 7.
    Polikar R (2012) Ensemble learning. Springer, USGoogle Scholar
  8. 8.
    Zhang CX, Duin RPT (2009) An empirical study of a linear regression combiner on multi-class data sets. In: Benediktsson, JA, Kittler, J, Roli, F (edn). Multiple classifier systems, proceedings, vol 5519. Lecture Notes in Computer Science, pp 478–487Google Scholar
  9. 9.
    Zhang ZL, Luo XG, Garcia S, Tang JF, Herrera F (2017) Exploring the effectiveness of dynamic ensemble selection in the one-versus-one scheme. Knowl-Based Syst 125:53–63Google Scholar
  10. 10.
    Zhu Y Q, Ou J S, Chen G, Yu H P (2011) Dynamic weighting ensemble classifiers based on cross-validation. Neural Comput Appl 20(3):309–317Google Scholar
  11. 11.
    Crook J N, Edelman D B, Thomas L C (2007) Recent developments in consumer credit risk assessment. Eur J Oper Res 183(3):1447–1465. MathSciNetzbMATHGoogle Scholar
  12. 12.
    Beque A, Coussement K, Gayler R, Lessmann S (2017) Approaches for credit scorecard calibration: an empirical analysis. Knowl-Based Syst 134:213–227. Google Scholar
  13. 13.
    Dietterich T G (2000) An experimental comparison of three methods for constructing ensembles of decision trees: bagging, boosting, and randomization. Mach Learn 40(2):139–157. Google Scholar
  14. 14.
    Fisher R A (1936) The use of multiple measurements in taxonomic problems. Ann Eugen 7(2):179–188Google Scholar
  15. 15.
    Hand D J, Henley W E (1997) Statistical classification methods in consumer credit scoring: a review. J Royal Stat Soc Ser A (Statistics in Society) 160:523–541Google Scholar
  16. 16.
    Marques A, García V, Sanchez J (2012) A literature review on the application of evolutionary computing to credit scoring. J Oper Res Soc 64(9):1384–1399Google Scholar
  17. 17.
    Tsai C -F, Chen M -L (2010) Credit rating by hybrid machine learning techniques. Appl Soft Comput 10 (2):374–380Google Scholar
  18. 18.
    Qian B, Rasheed K (2010) Foreign exchange market prediction with multiple classifiers. J Forecasting 29 (3):271–284. MathSciNetzbMATHGoogle Scholar
  19. 19.
    Sun J, Li H (2012) Financial distress prediction using support vector machines: ensemble vs. individual. Appl Soft Comput 12(8):2254–2265Google Scholar
  20. 20.
    Chen N, Ribeiro B, Chen A (2016) Financial credit risk assessment: a recent review. Artif Intell Rev 45(1):1–23Google Scholar
  21. 21.
    Li H, Sun J (2013) Predicting business failure using an RSF-based case-based reasoning ensemble forecasting method. J Forecasting 32(2):180–192MathSciNetzbMATHGoogle Scholar
  22. 22.
    Yu L A, Zhao Y, Tang L (2017) Ensemble forecasting for complex time series using sparse representation and neural networks. J Forecasting 36(2):122–138MathSciNetzbMATHGoogle Scholar
  23. 23.
    Zhou L G, Lu D, Fujita H (2015) The performance of corporate financial distress prediction models with features selection guided by domain knowledge and data mining approaches. Knowl-Based Syst 85:52–61. Google Scholar
  24. 24.
    Zhang C X, Duin R P W (2009) An empirical study of a linear regression combiner on multi-class data sets. In: Proceedings of multiple classifier systems, international workshop, MCS, vol 2009. Reykjavik, Iceland, pp 478–487Google Scholar
  25. 25.
    Abellán J, Mantas C J (2014) Improving experimental studies about ensembles of classifiers for bankruptcy prediction and credit scoring. Expert Syst Appl 41(8):3825–3830Google Scholar
  26. 26.
    Ala’raj M, Abbod M F (2016) Classifiers consensus system approach for credit scoring. Knowl-Based Syst 104:89–105. Google Scholar
  27. 27.
    Kim E, Kim W, Lee Y (2003) Combination of multiple classifiers for the customer’s purchase behavior prediction. Decis Support Syst 34(2):167–175Google Scholar
  28. 28.
    Zhang C X, Duin R P W (2011) An experimental study of one- and two-level classifier fusion for different sample sizes. Pattern Recogn Lett 32(14):1756–1767Google Scholar
  29. 29.
    Duin RPW, Tax DMJ (1998) Classifier conditional posterior probabilities. In: Joint Iapr international workshops on advances in pattern recognition, pp 611–619Google Scholar
  30. 30.
    Ting K M, Witten I H (1999) Issues in stacked generalization. J Artif Intell Res 10:271–289zbMATHGoogle Scholar
  31. 31.
    Kuncheva LI (2014) Combining pattern classifiers: methods and algorithms, 2nd ednGoogle Scholar
  32. 32.
    Yu L A, Yue W Y, Wang S Y, Lai K K (2010) Support vector machine based multiagent ensemble learning for credit risk evaluation. Expert Syst Appl 37(2):1351–1360Google Scholar
  33. 33.
    Jurek A, Bi Y X, Wu S L, Nugent C (2014) A survey of commonly used ensemble-based classification techniques. Knowl Eng Rev 29(5):551–581Google Scholar
  34. 34.
    Zhang L, Zhang L L, Teng W L, Chen Y B (2013) Based on information fusion technique with data mining in the application of finance early-warning. Procedia Comput Sci 17:695–703. Google Scholar
  35. 35.
    Ko A H R, Sabourin R, Britto A S (2008) From dynamic classifier selection to dynamic ensemble selection. Pattern Recogn 41(5):1718–1731zbMATHGoogle Scholar
  36. 36.
    Woloszynski T, Kurzynski M, Podsiadlo P, Stachowiak G W (2012) A measure of competence based on random classification for dynamic ensemble selection. Inf Fusion 13(3):207–213. Google Scholar
  37. 37.
    Woloszynski T, Kurzynski M (2011) A probabilistic model of classifier competence for dynamic ensemble selection. Pattern Recogn 44(10–11):2656–2668. zbMATHGoogle Scholar
  38. 38.
    Dos Santos E M, Sabourin R, Maupin P (2008) A dynamic overproduce-and-choose strategy for the selection of classifier ensembles. Pattern Recogn 41(10):2993–3009. zbMATHGoogle Scholar
  39. 39.
    Sun J, Fujita H, Chen P, Li H (2017) Dynamic financial distress prediction with concept drift based on time weighting combined with adaboost support vector machine ensemble. Knowl-Based Syst 120:4–14. Google Scholar
  40. 40.
    Cinlar E (2015) Introduction to stochastic process. IEEE Trans Syst Man Cybern SMC 3(5):533–533Google Scholar
  41. 41.
    Jarrow R A, Lando D, Turnbull S M (1997) A Markov model for the term structure of credit risk spreads. Rev Financ Stud 10(2):481–523Google Scholar
  42. 42.
    Timofeeva G A F, Timofeev N (2012) Forecasting credit portfolio components with a Markov chain model. Autom Remote Control 73(4):637–651MathSciNetzbMATHGoogle Scholar
  43. 43.
    Liu K, Lai KK, Guu S-M (2009) Dynamic credit scoring on consumer behavior using fuzzy Markov model. In: Fourth international multi-conference on computing in the global information technology, 2009. ICCGI’09. IEEE, pp 235–239Google Scholar
  44. 44.
    Fung E S, Siu T K (2012) A flexible Markov chain approach for multivariate credit ratings. Comput Econ 39(2):135–143zbMATHGoogle Scholar
  45. 45.
    Chen Y -K (2007) Economic design of variable sampling interval T 2 control charts—a hybrid Markov chain approach with genetic algorithms. Expert Syst Appl 33(3):683–689Google Scholar
  46. 46.
    Sousa M R, Gama J, Brandao E (2016) Dynamic credit score modeling with short-term and long-term memories: the case of Freddie Mac’s database. J Risk Model Validat 10(1):59–80Google Scholar
  47. 47.
    So M M C, Thomas L C (2011) Modelling the profitability of credit cards by Markov decision processes. Eur J Oper Res 212(1):123–130Google Scholar
  48. 48.
    Lipton A, Rennie A, Bielelcki T R, Crépey S, Herbertsson A (2012) Markov Chain models of portfolio credit risk. The Oxford Handbook of Credit Derivatives, Oxford. Google Scholar
  49. 49.
    Abdou H, Pointon J, El-Masry A (2008) Neural nets versus conventional techniques in credit scoring in Egyptian banking. Expert Syst Appl 35(3):1275–1292. Google Scholar
  50. 50.
    Vapnik V N (1995) The nature of statistical learning theory. IEEE Trans Neural Netw 8(6):988–999zbMATHGoogle Scholar
  51. 51.
    Zhou L G, Si Y W, Fujita H (2017) Predicting the listing statuses of Chinese-listed companies using decision trees combined with an improved filter feature selection method. Knowl-Based Syst 128:93–101. Google Scholar
  52. 52.
    Xu W, Xiao Z, Dang X, Yang D L, Yang X L (2014) Financial ratio selection for business failure prediction using soft set theory. Knowl-Based Syst 63:59–67. Google Scholar
  53. 53.
    Xu W, Xiao Z, Yang D L, Yang X L (2015) A novel nonlinear integrated forecasting model of logistic regression and support vector machine for business failure prediction with all sample sizes. J Test Eval 43(3):13. Google Scholar
  54. 54.
    UCI Machine Learning Repository (2013) University of California, School of Information and Computer Science.
  55. 55.
    Thomas L C, Crook J, Edelman D (2002), Credit scoring and its applications. SIAMGoogle Scholar
  56. 56.
    Yeh I C, Lien C H (2009) The comparisons of data mining techniques for the predictive accuracy of probability of default of credit card clients. Expert Syst Appl 36(2):2473–2480Google Scholar
  57. 57.
    Xiao H S, Xiao Z, Wang Y (2016) Ensemble classification based on supervised clustering for credit scoring. Appl Soft Comput 43:73–86. Google Scholar
  58. 58.
    Calabrese R, Osmetti S A (2015) Improving forecast of binary rare events data: a GAM-based approach. J Forecasting 34(3):230– 239MathSciNetzbMATHGoogle Scholar
  59. 59.
    Akkoc S (2012) An empirical comparison of conventional techniques, neural networks and the three stage hybrid adaptive neuro fuzzy inference system (ANFIS) model for credit scoring analysis: the case of Turkish credit card data. Eur J Oper Res 222(1):168–178. Google Scholar
  60. 60.
    Teng G -E, He C -Z, Xiao J, Jiang X -Y (2013) Customer credit scoring based on HMM/GMDH hybrid model. Knowl Inf Syst 36(3):731–747Google Scholar
  61. 61.
    Hand D J (2009) Measuring classifier performance: a coherent alternative to the area under the ROC curve. Mach Learn 77(1):103–123Google Scholar
  62. 62.
    Hand D J, Anagnostopoulos C (2013) When is the area under the receiver operating characteristic curve an appropriate measure of classifier performance? Pattern Recogn Lett 34(5):492–495Google Scholar
  63. 63.
    Garcia V, Marques A I, Sanchez J S (2015) An insight into the experimental design for credit risk and corporate bankruptcy prediction systems. J Intell Inf Syst 44(1):159–189Google Scholar
  64. 64.
    Demšar J (2006) Statistical comparisons of classifiers over multiple data sets. J Mach Learn Res 7:1–30MathSciNetzbMATHGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.School of Economics and Business AdministrationChongqing UniversityChongqingChina
  2. 2.College of Mathematics and StatisticsChongqing UniversityChongqingChina
  3. 3.School of Management Science and EngineeringShanxi University of Finance and EconomicsTaiyuanChina

Personalised recommendations