A LEAST SQUARES ALGORITHM FOR FITTING PIECEWISE LINEAR FUNCTIONS ON FIXED DOMAINS

Authors
  1. Trenholm, B.A.
Corporate Authors
Defence Research Establishment Atlantic, Dartmouth NS (CAN)
Abstract
A least squared error algorithm is presented for fitting a piecewise linear function to observed data, when the slope of the function is allowed to change only at specified points. The algorithm can be used to estimate piecewise linear trends in data, where the locations of possible changes in trend are known. Furthermore it can be used to reduce large quantities of data to manageable sizes, since J coordinates are input and L less than or equal to J coordinates are output. In addition, the algorithm is robust to problems of data dropout provided it is used cautiously. An example of a possible application is the fitting of a linear segment bathythermal profile (temperature vs. depth) to a large quantity of data.
Report Number
DREA-TM-85-215 — Technical Memorandum
Date of publication
15 Sep 1985
Number of Pages
37
DSTKIM No
85-03733
CANDIS No
48492
Format(s):
Hardcopy;Originator's fiche received by DSIS

Permanent link

Document 1 of 1

Date modified: