Abstract
In this paper, we illustrate how data envelopment analysis (DEA) can be used to aid interactive classification. We assume that the scoring function for the classification problem is known. We use DEA to identify difficult to classify cases from a database and present them to the decision-maker one at a time. The decision-maker assigns a class to the presented case and based on the decision-maker class assignment, a tradeoff cutting plane is drawn using the scoring function and decision-maker’s input. The procedure continues for finite number of iterations and terminates with the final discriminant function. We also show how a hybrid DEA and mathematical programming approach can be used when user interaction is not desired. For non-interactive case, we compare a hybrid DEA and mathematical programming based approach with several statistical and machine learning approaches, and show that the hybrid approach provides competitive performance when compared to the other machine learning approaches.
Similar content being viewed by others
Notes
More specifically, this is a basic BCC (Banker et al. 1984) input minimizing model with multiple inputs and one output that takes a constant value of unity for all accept class examples. The vector x is assumed to represent m-dimensional inputs. For BCC input model please see http://www.deazone.com and select BCC input under model tab.
This is a basic BCC output maximizing model with multiple outputs and one input that takes a constant value of unity for all reject class examples. The vector x is assumed to represent m-dimensional outputs. For BCC output model please see http://www.deazone.com and select BCC output under model tab.
Because such a line will be below the accept class frontier.
Because such a line will be above the reject class frontier.
References
Altman, E. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance, 23, 589–609.
Asparoukhov, O. K., & Stam, A. (1997). Mathematical programming formulations for two-group classification with binary variables. Annals of Operations Research, 74, 89–112.
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in DEA. Management Science, 20(9), 1078–1092.
Barber, C. B., Dobkin, D. P., & Huhdanpaa, H. T. (1996). The quickhull algorithm for convex hulls. ACM Transactions on Mathematical Software, 22, 469–483.
Bhattacharyya, S., & Pendharkar, P. C. (1998). Inductive, evolutionary and neural techniques for discrimination: A comparative study. Decision Sciences, 29(4), 871–900.
Feinberg, F. M., & Huber, J. (1996). A theory of cutoff formation under imperfect information. Management Science, 42(1), 65–84.
Gaba, A., & Viscusi, W. K. (1998). Differences in subjective risk thresholds: Worker groups as an example. Management Science, 44(6), 801–811.
Gallagher, R. J., Lee, E. K., & Patterson, D. A. (1997). Constrained discriminant analysis via 0/1 mixed integer programming. Annals of Operations Research, 74, 65–88.
Geoffrion, A. M., Dyer, J. S., & Feinberg, A. (1972). An interactive approach for multicriterion optimization, with an application to the operation of an academic department. Management Science, 19, 357–368.
Huang, Z., & Li, S. X. (2001). Stochastic DEA models with different types of input-output disturbances. Journal of Productivity Analysis, 15, 95–113.
Ohlson, J. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 19, 109–131.
Pavur, R., Wanarat, P., & Loucopoulos, C. (1997). Examination of the classificatory performance of MIP models with secondary goals for the two-group discriminant problem. Annals of Operations Research, 74, 173–189.
Pedro Duarte Silva, A., & Stam, A. (1997). A mixed integer programming algorithm for minimizing the training sample misclassification cost in two-group classification. Annals of Operations Research, 74, 129–157.
Pendharkar, P. C. (2002). A potential use of DEA for inverse classification problem. Omega: An International Journal of Management Science, 30, 243–248.
Pendharkar, P. C. (2011). A hybrid radial basis function and data envelopment analysis neural network for classification. Computers & Operations Research, 38(1), 256–266.
Pendharkar, P. C., Subramanian, G. H., & Rodger, J. A. (2005). A probabilistic model for predicting software development effort. IEEE Transactions on Software Engineering, 31(7), 615–624.
Rubin, P. A. (1997). Solving mixed integer classification problems by decomposition. Annals of Operations Research, 74, 51–64.
Seiford, L. M., & Zhu, J. (1998). An acceptance system decision rule with data envelopment analysis. Computers & Operations Research, 25(4), 329–332.
Shin, W. S., & Ravindran, A. (1991). An interactive method for multi-objective mathematical programming problems. Journal of Optimization Theory and Applications, 68, 539–561.
Troutt, M. D. (1994). Direction-specific gradient scaling for interactive multicriterion optimization using an abstract mass concept. Operations Research, 42(6), 1110–1119.
Troutt, M. D. (1995). A maximum decisional efficiency estimation principle. Management Science, 41(1), 76–82.
Troutt, M. D., Rai, A., & Zhang, A. (1996). The potential use of DEA for credit applicant acceptance systems. Computers & Operations Research, 23(4), 405–408.
Troutt, M. D., Rai, A., & Tadisina, S. K. (1997). Aggregating multiple expert data using the maximum decisional efficiency principle. Decision Support Systems, 21, 75–82.
Zahedi, F. (1993). Intelligent systems for business: expert systems with neural networks. Belmont: Wadsworth.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Pendharkar, P.C., Troutt, M.D. Interactive classification using data envelopment analysis. Ann Oper Res 214, 125–141 (2014). https://doi.org/10.1007/s10479-012-1091-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10479-012-1091-8