Influence in Testing the Equality of Two Covariance Matrices

두개의 공분산 행렬의 동질성 검정에서의 영향치 분석

  • Myung Geun Kim (Department of Applied Statistics, Seowon University, 360-742, Korea)
  • Published : 1994.09.01

Abstract

A diagnostic method useful for detecting outliers in testing the equality of two covariance metrics is developed using the influence curve approach. This method is easily generalized to more than two covariance matrices. A sample version for the influence measure of detecting outliers is considered based on the empirical distribution functions. The sample version includes as its component terms the well-known test statistic for detecting one outlier at a time introduced by Wilks and its generalization to the two-group case.

두개의 공분산 행렬의 동질성을 검정하는데 있어서, influence curve 방법을 이용하여 outlier를 찾는데 유용한 진단법을 제시한다. 이러한 진단법은 두개 이상의 공분산 행렬의 경우에 쉽게 일반화된다. 경험적 분포함수에 입각한 진단법의 sample version을 고려하며, 이것은 Wilks가 제안한 한개의 outlier를 찾는데 필요한 통계량과 두개의 모집단의 경우로 일반화된 Wilks 통계량을 포함한다.

Keywords

References

  1. Applied Statistics v.36 A New Graphical Method for Detecting Single and Multiple Outliers in Univariate and Multivariate Data Bacon-Shone,J.;W.K.Fung
  2. Applied Statistics v.27 The Influence Function as an Aid in Outlier Detection in Discriminant Analsis Campbell,N.A.
  3. Applied Statistics v.41 Sequential Application of Wilks's Multivariate Outlier Test Caroni,C.;Prescott,P.
  4. Residuals and Influence in Regression Cook,R.D.;Weisberg,S.
  5. Biometrika v.72 Influence in Principal Component Analysis Critchley,F.
  6. Journal of the American Statistical Association v.20 The Influence Curve and its Role in Robust Estimation Hampel,F.R.
  7. Applied Multivariate Statistical Analysis Johnson,R.A.;Wichern,D.W.
  8. Observations Seber,G.A.F.
  9. Sankhya v.25 Multivariate Statistical Outliers Wilks,S.S.