Re: DM: Applications of Canonical Correlations in Data Mining

[Warren Sarle]

Krishnadas writes:
> I would like to know if there have been any application of canonical
> correlations in datamining problems. Would appreciate pointers to 
> papers/references, software, success stories.

Seems unlikely. Data mining applications are usually predictive, but
canonical correlation does not try to predict any of the observed
variables. You can have a large canonical correlation (meaning that 
some
linear combination can be predicted well) but low correlations between
canonical variables and the observed variables (meaning that the
observed variables cannot be predicted well). Often when people do
canonical correlation, they really should have done maximum redundancy
analysis (AKA principal components of instrumental variables), which 
is
a predictive dimensionality-reduction method for multiple independent
and multiple dependent variables.

   Fortier, J.J. (1966), "Simultaneous Linear Prediction,"
   Psychometrika, 31, 369-381.

   Rao, C.R. (1964), "The Use and Interpretation of Principal
   Component Analysis in Applied Research," Sankya A, 26, 329-358.

   van den Wollenberg, A.L. (1977), "Redundancy Analysis--An
   Alternative to Canonical Correlation Analysis," Psychometrika,
   42, 207-219.


-- 

Warren S. Sarle       SAS Institute Inc.   The opinions expressed here
saswss@unx.sas.com    SAS Campus Drive     are mine and not 
necessarily
(919) 677-8000        Cary, NC 27513, USA  those of SAS Institute.