Hello ROOT experts,
I am currently working with TMVA for a regression analysis with a substantial number of variables. I’d like to perform a Principal Component Analysis (PCA) to account for potential correlations among these variables, using the option VarTransform=P.
Regrettably, I encountered this message:
Factory : Train method: LD for Regression
:
: Preparing the Principle Component (PCA) transformation...
: ----------------------------------------------------------------------------
: : More than 200 variables, will not calculate PCA!
: ----------------------------------------------------------------------------
Then the process results in a segmentation violation error.
I read that the transformation is facilitated through the TPrincipal library, which typically manages a significant volume of variables without issue. I am curious if there exists a global configuration option in TMVA that can be adjusted to circumvent this warning and enable the PCA calculation despite the large number of variables.
Many thanks.