Correlated Variables in PCA



The above article says “correlated variables shouldbe removed for PCA, as PCA places more importance on the correlated variables”. But whereas other article

and many other articles say that PCA is used when we come across correlated varibles. So should I remove correlated varibles before PCA or don’t remove them?


Hi @raka747,

Generally, PCA performs better when we have uncorrelated variables. So, it is a good practice to remove correlated variables before applying PCA.

I would suggest you to apply PCA twice. First on the data having correlated variables and secondly after removing the correlated variables. Then you can visualize it more effectively. Also, do share your results that will be helpful for the community.