Some Integrated Squared Error Procedures for Multivariate Normal Data,

Abstract

Two methods of estimation for the parameters of the multivariate normal distribution based on the sample characteristic function are given. These methods are shown to have an equivalent basis in terms of Parzen kernel-like density estimation. The estimators for the mean vector and covariance matrix are dependent on a user-specified parameter. Variation of the user-specified parameter produces a response surface in the parameter estimates and therefore allows for an informal sensitivity analysis of the data with respect to a tentative working model. The informal sensitivity analysis is intricately related to formal tests of fit of the model. The estimators of mean vector and covariance matrix have desirable robustness properties, are easy to compute and use, are relatively efficient at the multivariate normal, and are useful in identifying potential outliers and inconsistencies in some statistical assumptions. These methods are directly applicable to structured data such as multivariate experimental designs. Several illustrations are provided.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1986
Accession Number
ADA178985

Entities

People

  • A. S. Paulson
  • C. E. Lawrence
  • H. L. Hwang
  • N. J. Delaney

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Chemical Analysis
  • Computations
  • Covariance
  • Data Science
  • Efficiency
  • Equations
  • Estimators
  • Experimental Design
  • Gaussian Distributions
  • Information Science
  • Normal Distribution
  • Observation
  • Probability
  • Sensitivity
  • Statistical Algorithms
  • Statistical Samples

Fields of Study

  • Mathematics

Readers

  • Statistical inference.
  • Systems Analysis and Design