Hi,
Anyone familiar with projection pursuit? I have a dataset of 250
variables and 78 observations, I would like to reduce the dimensionality and
use components in my further classification analysis. Here, I have a
question.
Should I project the data in all 78 dimensions and choose few components
(just like PCA)? or should I project the data in lower dimension (say 2 or
3) and use all the components?
Any help would be appreciated.
Arun
--
View this message in context: http://www.nabble.com/Projection-Pursuition-tp20806435p20806435.html
Sent from the AllStat mailing list archive at Nabble.com.
|