I have recently learnt PCA and was unsure about how to explain to business about the Principal Components when used in a model.

As we all know PCA is used for dimensionality reduction. After PCA you get orthogonal principal components that capture maximum variation using least possible components. Each component thus obtained is a linear combination of original variables. which original component is governed by which original variables can be easily found out by looking at the loading factor. One can then use these components for linear models like linear or logistic regression.

If I use PCA for dimensionality reduction before linear regression, the model output of linear regression is in the form of principal components instead of original variables. How do I explain these to business. The original variables are very well explainable but these derived components cannot be explained plainly. I can say the house of a price depends on the area of the house. But it;s difficult to explain that it is some linear combination of area of house, no of floors, locality, etc. to business.

Similar questions and discussions