Testing for Periodicity in a Time Series.

Abstract

In 1929, Sir R. A. Fisher proposed a test for periodicity in a time series based on the maximum spectrogram ordinate. This paper proposes a one-parameter family of tests that contains Fisher's test as a special case. It is shown how to select from this family a test that will have substantially larger power than Fisher's test against many alternatives, yet will lose only negligible power against alternatives for which Fisher's test is known to be optimal. Critical values are calculated and tabled using a duality with the problem of covering a circle with random arcs. The power is studied using Monte Carlo techniques. The method is applied to the study of the magnitude of a variable star, showing that these power gains can be realized in practice.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1978
Accession Number
ADA060657

Entities

People

  • Andrew F. Siegel

Organizations

  • University of Wisconsin–Madison

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Health
  • Health Services
  • Mathematics
  • Military Research
  • North Carolina
  • Periodic Variations
  • Plastic Explosives
  • Power Gain
  • Probability
  • Public Health
  • Stars
  • Statistics
  • United States
  • Variable Stars
  • White Noise
  • Wisconsin

Fields of Study

  • Mathematics

Readers

  • Regression Analysis.