Principal component regression, keeping variable names
8 views (last 30 days)
Show older comments
Hi everyone,
I am performing a principal component analysis in Matlab and I have a problem: as I understand Matlab algorithm for SVD and PCA returns sorted values for eigenvalues and the corresponding eigenvectors, so making it impossible to identify the original data. Since I want a visual formula of my regression to show for an important project and the predictors are at least ten I need to find a way to keep the variable name on each array in order to being able to identify the variables at the end of the principle component regression. I am grateful for anyone who can help in any way.
Diletta
0 Comments
Answers (1)
Bjorn Gustavsson
on 16 Sep 2021
Well the eigen-vectors will correspond to some linear combinations of your orignal variables. As such one eigenvector might for example correspond to [0.267*apples, 0.534*oranges, 0.801*pigs/m^2] - depending on what your original variables are...
0 Comments
See Also
Categories
Find more on Dimensionality Reduction and Feature Extraction in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!