PREDICTION OF TIME SERIES USING MULTIPLE REGRESSION TECHNIQUES AND SEAKEEPING APPLICATIONS
Abstract
This report discusses methods for predicting future values of discrete time series from past observed values of the time series. The points at which the autocorrelation function is computed are the past points used in the analysis. Classical ideas are reviewed, and then extended to handle more advanced time series problems. Applications of these results are explained for seakeeping applications of: (1) long range ocean prediction, (2) short term roll prediction, and (3) vibration response prediction. Further material appears on statistical tests for coefficient determination, and on digital computer requirements.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1963
- Accession Number
- AD0432475
Entities
People
- Loren D. Enochson