Journal of Intelligence and Information Systems (지능정보연구)

Korea Intelligent Information System Society (한국지능정보시스템학회)
권호 표지

Aggregating Prediction Outputs of Multiple Classification Techniques Using Mixed Integer Programming

다수의 분류 기법의 예측 결과를 결합하기 위한 혼합 정수 계획법의 사용

  • Jo, Hongkyu (Financial Engineering Research Center, NICE Pricing Services) ;
  • Han, Ingoo (Graduate School of Management, Korea Advanced Institute of Science and Technology)
  • Published : 2003.06.01
Download PDF
⟨ Previous Next ⟩

Abstract

Although many studies demonstrate that one technique outperforms the others for a given data set, there is often no way to tell a priori which of these techniques will be most effective in the classification problems. Alternatively, it has been suggested that a better approach to classification problem might be to integrate several different forecasting techniques. This study proposes the linearly combining methodology of different classification techniques. The methodology is developed to find the optimal combining weight and compute the weighted-average of different techniques' outputs. The proposed methodology is represented as the form of mixed integer programming. The objective function of proposed combining methodology is to minimize total misclassification cost which is the weighted-sum of two types of misclassification. To simplify the problem solving process, cutoff value is fixed and threshold function is removed. The form of mixed integer programming is solved with the branch and bound methods. The result showed that proposed methodology classified more accurately than any of techniques individually did. It is confirmed that Proposed methodology Predicts significantly better than individual techniques and the other combining methods.

경영 분류 문제에 대한 많은 연구들은 여러가지 기법들간의 성과 비교에 대한 것이었지만, 각각의 연구들마다 가장 좋은 기법이 어떤 것인가에 대해서는 상이한 결론을 내고 있다. 다수의 분류 기법 중에서 가장 좋은 것을 사용하는 방법에 대한 대안으로,분류 기법을 통합하여 성과를 향상시키는 방법이 있다. 본 연구에서는 개별 분류 기법의 결과를 선형 결합하여 예측력을 높이는 방법을 제시하였다. 최 적 선형 결합 가중치를 계산하기 위해 혼합 정수 계 획 법을 사용하였다. 목적 함수로 사용한 오분류 비용의 최소화에서 오분류 비용은 부도 기업을 모형에서 정상으로 예측한 오류와 정상기업을 모형에서 부도 기업으로 예측한 오류의 합으로 정의하였다. 문제 풀이 과정을 단순화하기 위하여 본 논문에서는 절사점 (cutoff value)을 고정하였고, 경계 함수 (threshold function)를 배제하였다. 정수계획법의 계산을 위해 branch 8, bound 방법을 사용하였다. 선형 결합에 의한 모형의 예측력이 개별 기법에 의해 구축된 모형의 예측력을 상회하였고, 그 차이가 통계적으로도 유의하였다.

Keywords

References

  1. Journal of Finance v.23 no.3 Financial ratios, discriminant analysis, the prediction of corporate bankruptcy Altman,E.
  2. Journal of Banking and Finance v.18 Corporate distress diagnosis: Comparisons using linear discriminant analysis and neural networks (the Italian experience) Altman,E.;G.Marco;F.Varetto
  3. Journal of Accounting Research v.5 no.Supp. Financial ratios as predictors of failure. Empirical research in accounting: Selected studies Beaver,W.
  4. Decision Science v.18 Predicting stock market behavior through rule-induction: An application of the learning-from-example approach Braun,H.;J.Chandler
  5. Proceedings of the $1^{st}$ International Conference on Artificial Intelligence Applications on Wall Street Neural networks and the mathematics of chaos - An investigation of these methodologies as accurate predictors of corporate bankruptcy Cadden,D.
  6. Intelligent Systems in Accounting, Finance, & Management v.3 A comparative analysis of artificial neural networks using financial distress prediction Fanning,K.M.;K.O.Cogger
  7. Journal of American Taxation Association Symbolic concept acquisition: A new approach to determining underlying tax law construct Garrison,L.;R.Michaelson
  8. Intelligent Systems in Accounting, Finance and Management v.2 Performance of neural networks in managerial forecasting Jhee,W.C.;J.K.Lee
  9. Expert Systems With Applications v.11 no.4 Interation of case-based forecasting, neural network, and discriminant analysis for bankruptcy prediction Jo,H.;I.Han
  10. Expert Systems With Applications v.13 no.2 Bankruptcy prediction using case-base reasoning, neural networks, and discriminant analysis Jo,H.;I.Han;H.Lee
  11. Working Paper Hybrid classifiers for knowledge discovery Kumar,A.;I.Ormeda
  12. Expert Systems With Applications v.1 A comparative study of recursive partitioning algorithm and analog concept learning system Lee,S.B.;S.H.Oh
  13. Expert Systems With Applications v.1 Integrating statistical and inductive learning methods for knowledge acquisition Liang,T.P.;J.S.Chandler;I.Han
  14. European Journal of Operational Research v.138 Genetic programming and rough sets: A hybrid approach to bankruptcy classifiation McKee,T.E.;T.Lensberg
  15. Decision Sciences v.26 Combining neural networks and statistical predictions to solve the classification problem in discriminant analysis Markham,I.S.;C.T.Ragsdale
  16. Neural Networks in Finance and Investing A neural network model for bankruptcy prediction Odam,M.;R.Sharda;R.Trippi(ed.);E.Turban(ed.)
  17. Expert Systems With Applications v.23 no.3 A case-based reasoning with the feature weights derived by analytic hierarchy process for bankruptcy prediction Park,C.;I.Han
  18. Neural Networks in Finance and Investing A neural network approach to bankruptcy prediction Raghupathi,W.;L.Schkade;B.Raju;R.Trippi(ed.);E.Turban(ed.)
  19. Expert Systems With Applications v.23 no.3 A genetic algorithm appliation in bankruptcy prediction modeling Shin,K.;Y.Lee
  20. Proceedings of the IEEEE Conference on Neural Networks v.Ⅱ Neural neworks for ond ratig improved by multiple hidden layers Surkan,A.;J.Singleon
  21. Management Science v.38 Managerial applications of neural networks: The case of bank failure predictions Tam,K.;M.Kiang
  22. Accounting, Management, & Information Technology v.4 no.3 Incorporating complementary ratios in the analysis of financial statements Trigueiros,D.
  23. Decision Support Systems v.21 Aggregating multiple expert data for linear case valuation models using the MDE principle Troutt,M.D.;A.Rai;S.K.Tadisina