On EM Algorithm For Discrete Classification With Bahadur Model: Unknown Prior Case

  • Kim, Hea-Jung (Department of Statistics, Dongguk University, Seoul 100-715) ;
  • Jung, Hun-Jo (Department of Computer Science and Statistics, Hanseo University, Chungnam 352-820)
  • 발행 : 1994.06.01

초록

For discrimination with binary variables, reformulated full and first order Bahadur model with incomplete observations are presented. This allows prior probabilities associated with multiple population to be estimated for the sample-based classification rule. The EM algorithm is adopted to provided the maximum likelihood estimates of the parameters of interest. Some experiences with the models are evaluated and discussed.

키워드

참고문헌

  1. An Introduction to Multivariate Analysis Anderson,T.W.
  2. Studies in Item Analysis and Prediction A representation of the joint distribution of response to n dichotomous items Bahadur,R.R.;H.Solomon(ed.)
  3. Journal of Royal Statistical Society v.B39 Maximum likelihood estimation from incomplete data via the EM algorithm(with discussion) Dempster,A.P.;Laird,N.M.;Rubin,D.B.
  4. Discrete Discriminant Analysis Dillon,W.R.
  5. Journal of the American Statistical Association v.73 On the performance of some multinomial classification rules Dillon,W.R.;Goldstein,M.
  6. Journal of Royal Statistical Society, B v.26 Posterior odds for multivariate normal classifications Geisser,S.
  7. Testing Statistical Hypothesis Lehmann,E.L.
  8. Journal of the American Statistical Association v.68 Evaluation of five discriminant procedures for binary variables Moore,D.H.
  9. Studies in Item Analysis and Prediction Solomon,H.(ed.)
  10. Scandinavian Journal of Statistics v.1 Maximum likelihood theory of incomplete data from an exponential family Sundberg,R.
  11. Annals of Statistics v.11 On convergence properties of the EM algorithm Wu,C.F.J.