PCA total variable contribtionI would like to get a list of variables ordered by their contribution explaining the variability within the data.
In PCA, one can plot the partial contribution of the variables. However, this visualization does not consider the different importance between PCs. The eigenvalues table includes the percentage of data explained.
Is there a way to get the total variable contribution?
AnalyticsScriptingRe: PCA total variable contribtionHi @FN,
I think you are looking for loadings, or the eigenvectors * sqrt( eigenvalues ). These are calculated in the PLS platform within JMP but not PCA. Here is how I would go about that interactively for PCA:
Re: PCA total variable contribtionHi,
I'm not aware of a direct method to get to the Total Contribution values but here is a possible work around (not particularly elegant) that may do the job for you.
1) Save the Partial Contribution table (right click > make into table)
2) Transpose the resulting table
3) Save the EigenValues table into a new table
4) Copy the Percent column to your Partial Contribution table
5) Calculate the product of each partial contribution by the corresponding Eigenvalues
This is rather labor intensive so if you have to retrieve this information for multiple tables, you may want to look into scripting those steps.
Best,
TS
