Nonparametric Estimation of the Generalized Variance.
Abstract
For multivariate distributions with finite second order moments, a nonparametric symmetric, unbiased estimator of the generalized variance is considered, and it is shown to be (nonparametric) optimal for the class of distributions having finite fourth order moments. A jackknifed version of the sample generalized variance is also considered as a contender; it is computationally more convenient and asymptotically equivalent to the former. It is also shown that the second estimator performs quite well (in large sample) relative to the optimal normal theory estimators under several loss functions. (Keywords: kernels; U-statistics; von mises' functionals).
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1986
- Accession Number
- ADA186029
Entities
People
- Bimal K. Sinha
- Pranab K. Sen
Organizations
- University of Pittsburgh