Premium
Multivariate least squares and its relation to other multivariate techniques
Author(s) -
Lipovetsky Stan,
Tishler Asher,
Conklin W. Michael
Publication year - 2002
Publication title -
applied stochastic models in business and industry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.413
H-Index - 40
eISSN - 1526-4025
pISSN - 1524-1904
DOI - 10.1002/asmb.462
Subject(s) - multivariate statistics , multivariate analysis , relation (database) , partial least squares regression , canonical correlation , principal component analysis , least squares function approximation , computer science , statistics , mathematics , econometrics , data mining , estimator
We consider multivariate least squares (LS) for the estimation of the connection between two data sets and show how LS is related to other multivariate techniques regularly used to analyse large data sets. LS methods are shown to be equivalent, or similar, to principal components, canonical correlations and its modifications, including a variant of the partial LS. LS approach provides a convenient unified framework for a general description and comparison of various multivariate methods, facilitates their understanding, and helps to identify their usefulness for various real‐world applications. As an example we estimate and discuss the relations between data sets containing managerial variables and success measures. Copyright © 2002 John Wiley & Sons, Ltd.