DocumentCode
837160
Title
On reliability of prediction in linear models
Author
Ward, Rabab Kretdieh
Author_Institution
University of British Columbia, Vancouver, BC, Canada
Volume
26
Issue
6
fYear
1981
fDate
12/1/1981 12:00:00 AM
Firstpage
1297
Lastpage
1299
Abstract
For a linear noisy system assume an estimate 
of the unknown parameter vector was obtained from a certain past data on inputs and outputs of the system. The prediction of the output for any input vector 
is a linear combination of and . The question of concern is: For what - values of the inputs would the error in their predicted outputs be the smallest? It is found that they are the inputs which lie on the fitted plane of the orthogonal regression among the past data of the clean inputs.
of the unknown parameter vector was obtained from a certain past data on inputs and outputs of the system. The prediction of the output for any input vector is a linear combination of and . The question of concern is: For what - values of the inputs would the error in their predicted outputs be the smallest? It is found that they are the inputs which lie on the fitted plane of the orthogonal regression among the past data of the clean inputs.Keywords
Linear systems; Prediction methods; Automatic control; Estimation error; Filtering; Goniometers; Noise measurement; Notice of Violation; Phase change random access memory; Predictive models; Signal to noise ratio; Vectors;
fLanguage
English
Journal_Title
Automatic Control, IEEE Transactions on
Publisher
ieee
ISSN
0018-9286
Type
jour
DOI
10.1109/TAC.1981.1102797
Filename
1102797
Link To Document