Robust two-group discrimination by bounded influence regression. A Monte Carlo simulation. (English) Zbl 0937.62513

Summary: Since linear discriminant analysis (LDA) and multiple linear regression (MR) are numerically equivalent in the two-group case, robust regression can be used to devise a robust discriminant analysis. While \(M\)-estimators will be affected by outliers in the classification variables, \(GM\)-estimators will resist them. Monte Carlo simulation is used to evaluate the performance of several \(GM\)-estimators as applied to the problem of two-group discrimination, with respect to probability of misclassification. It is concluded that the \(GM\)-classifiers perform well, especially some of them. An alternative approach based on high breakdown point estimators of location and scatter is proposed for cases of heavy contamination, when the performance of \(GM\)-estimators breaks down.


62-XX Statistics


Full Text: DOI


[1] Ahmed, S. W.; Lachenbruch, P. A.: Discriminant analysis when scale contamination is present in the initial sample. Classification and clustering, 331-353 (1977)
[2] Campbell, N. A.: Robust procedures in multivariate analysis I: Robust covariance estimation. Applied statistics 29, No. 3, 231-237 (1980) · Zbl 0471.62047
[3] Dutter, R.: BLINWDR: A computer program for robust and bounded influence regression. Statistical data analysis based on the L1-norm and related methods (1987) · Zbl 0494.62003
[4] Hettmansperger, T. P.: Why not try a robust regression. Australian journal of statistics 29, 1-18 (1987) · Zbl 0647.62065
[5] Hampel, F. R.; Ronchetti, E. M.; Rousseeuw, P. J.; Stahel, W. A.: Robust statistics. The approach based on influence functions. (1986) · Zbl 0593.62027
[6] Krasker, W. S.; Welsch, R. E.: Efficient bounded-influence regression estimation. Journal of the American statistical association 77, 595-604 (1982) · Zbl 0501.62062
[7] Krusinska, E.: Robust methods in discriminant analysis. Rivista di statistica applicada 21, No. 3, 239-253 (1988)
[8] Krusinska, E.; Liebhart, J.: Robust selection of the most discriminative variables in the dichotomous problem with application to some respiratory disease data. Biometrical journal 30, No. 2, 295-304 (1988)
[9] Lachenbruch, P. A.: Discriminant analysis. (1975) · Zbl 0354.62050
[10] Marazzi, A.: Solving bounded influence regression problem with ROBSYS. Statistical data analysis based on the L1-norm and related methods (1987)
[11] Maronna, R. A.; Bustos, O.; Yohai, V.: Bias- and efficiency-robustness of general M-estimators for regression with random carriers. Smoothing techniques for curve estimation, 91-116 (1979) · Zbl 0416.62050
[12] Neykov, N. M.; Neytchev, Pl.N.: Unmasking multivariate outliers and leverage points by means of BMDP3R and BMDPAR. Directions in robust statistics and diagnostics, 115-128 (1991)
[13] Neykov, N. M.; Neytchev, Pl.N.: Robust and bounded-influence regression using BMDP3R and BMDPAR. (1989)
[14] Rousseeuw, P. J.: Multivariate estimation with high breakdown point. Mathematical statistics and applications, 283-297 (1985) · Zbl 0609.62054
[15] Rousseeuw, P. J.; Van Zomeren, B. C.: Unmasking multivariate outliers and leverage points (with discussion). Journal of the American statistical association 85, 633-639 (1990)
[16] Samarov, P.; Welsch, R. E.: Computational procedure for bounded influence regression. COMPSTAT 1982: Proceedings in computational statistics (1982) · Zbl 0489.65080
[17] Todorov, V.: Computing the minimum volume determinant estimator (MCD) by simulated annealing. Computational statistics and data analysis 14 (1992) · Zbl 0937.62519
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.