THE PERIODIC ANALYSIS OF SAMPLED DATA
MASSACHUSETTS INST OF TECH LEXINGTON LINCOLN LAB
Pagination or Media Count:
A brief survey is given of some currently used techniques of periodic analysis of sampled data. Almost every periodic analysis of empirical data eventually relies on one of the following three techniques 1 regression analysis on a trigonometric polynomial 2 periodogram analysis 3 the estimation of power spectra. Each of these techniques is best applied under distinct modeling assumptions, and an attempt has been made to discuss their applicability and their range of validity under various circumstances. After a brief introduction in Chapter I, Chapter II proceeds to describe the models in use and the statistical properties of the estimates associated with them. The effects of sampling and some of their implications in practical situations are treated in Chapter III. Chapter IV relates several common question areas to the theory developed in the earlier portions of the report, and presents some illustrations of the power of the techniques discussed. Finally, Chapter V gives the briefest outline of the general requirements of a comprehensive program in time series data analysis.