Determination of Smoothing for Spectral-Matrix Estimation

Abstract

A statistical procedure for determining the resolution to be used in spectral-matrix estimation is needed for the exploratory analysis of non- stationary multiple time series. The procedure accomplished this by creating periodograms for a frequency-time interval and then successively dividing this interval into subintervals. At each step, the division is chosen by comparing spectral matrices using the largest-root statistic. This multiple-decision rule is shown to be consistent. Small-sample biases and approximate critical regions for testing stationarity and frequency uniformity are found. Simulation shows that simultaneous analysis of all the series performs better than individual analyses.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 26, 1973
Accession Number
AD0769922

Entities

People

  • Walter S. Liggett Jr.

Organizations

  • RTX

Tags

Communities of Interest

  • Air Platforms
  • Energy and Power Technologies
  • Ground and Sea Platforms

DTIC Thesaurus Topics

  • Algorithms
  • Commerce
  • Consistency
  • Covariance
  • Data Analysis
  • Data Science
  • Eigenvalues
  • Factor Analysis
  • Frequency
  • Frequency Bands
  • Information Science
  • Military Research
  • Multivariate Analysis
  • Rhode Island
  • Simulations
  • Statistics
  • Time Intervals

Fields of Study

  • Mathematics

Readers

  • Regression Analysis.
  • Statistical inference.