Resampling Methods for Error Rate Estimation in Discriminant Analysis

  • Masayuki Honda
  • Sadanori Konishi
Conference paper
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)


The performance of resampling methods, like the bootstrap or cross-validation methods, was investigated for estimating the error rates in linear and quadratic discriminant analyses. A Monte Carlo experiment was carried out under the assumption that two population distributions were characterized by a mixture of two multivariate normal distributions. Simulation results indicated that the bootstrap method gave good performance in the case of the linear discriminant function, but it was a little biased when the quadratic discriminant function was used. Cross-validation method was superior in regard to the unbiasedness, and the 0.632 bootstrap estimator outperformed in regard to the mean square error. The methods for error rate estimation were also examined through the analysis of real data in medical diagnosis.


Mean Square Error Linear Discriminant Analysis Discriminant Function Multivariate Normal Distribution Quadratic Discriminant Analysis 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. Andrews, D. F. and Herzberg, A. M. (1985): Data: A Collection of problems from many fields for the student and research worker. Springer-Verlag, New York.MATHGoogle Scholar
  2. Ashikaga, T. and Chang, P. C. (1981): Robustness of Fisher’s linear discriminant function under two-component mixed normal models. J. Amer. Statist. Assoc. 76, 676–680.MATHCrossRefGoogle Scholar
  3. Efron, B. (1979): Bootstrap methods: Another look at the jackknife. Ann. Statist. 7, 1–26.Google Scholar
  4. Efron, B. (1983): Estimating the error rate of a prediction rule: Improvement on cross-validation. J. Amer. Statist. Assoc. 78, 316–331.Google Scholar
  5. Fitzmaurice, G. M., Krzanowski, W. J. and Hand, D. J. (1991): A Monte Carlo study of the 632 bootstrap estimator of error rate. J. of Classification 8, 239–250.CrossRefGoogle Scholar
  6. Ganeshanandam, S and Krzanowski, W. J. (1990): Error-rate estimation in two-group discriminant analysis using the linear discriminant function. J. Statist. Comput. Simul. 36, 157–175.MathSciNetCrossRefGoogle Scholar
  7. Konishi, S. and Honda, M. (1990): Comparison of procedures for estimation of error rates in discriminant analysis under nonnormal populations. J. Statist. Comput. Simul. 36, 105–115.MathSciNetCrossRefGoogle Scholar
  8. Mardia, K. V. (1970): Measures of multivariate skewness and kurtosis with applications. Biometrika 57, 519–530.MathSciNetMATHCrossRefGoogle Scholar
  9. Marks, S and Dunn, O. J. (1974): Discriminant functions when covariance matrices are unequal. J. Amer. Statist. Assoc. 69, 555–559.MATHCrossRefGoogle Scholar
  10. McLachlan, G. J. (1974): An asymptotic unbiased technique for estimating the error rates in discriminant analysis. Biometrics 30, 239–249.MathSciNetMATHCrossRefGoogle Scholar
  11. Wahl, P and Kronmal, R. (1977): Discriminant functions when covariances are unequal and sample sizes are moderate. Biometrics 33, 479–484.MATHCrossRefGoogle Scholar

Copyright information

© Springer Japan 1998

Authors and Affiliations

  • Masayuki Honda
    • 1
  • Sadanori Konishi
    • 2
  1. 1.Division of Medical InformaticsChiba University HospitalChuou-ku Chiba 260Japan
  2. 2.Graduate School of MathematicsKyushu UniversityHigashi-ku Fukuoka 812Japan

Personalised recommendations