I am trying to understand the different ways in which PCA and Factor analysis create the factor loadings. Specifically speaking I was trying to see the difference in the rotated Principal components and the loadings of the factor analysis model.I had used PCA first and then tried using factanal on the same data:
churn.pca <- princomp(churn[,-15],center = T,scale = T) #Rotation: pca.rotate <- varimax(churn.pca$loadings) #Using factanal: churn.fa <- factanal(churn[,-15],factors = 5,rotation = "varimax")
However,PCA after rotation returned:
whereas factanal gave an error:
For the rotated PC’s we can say that the variable Total.Night.Minutes loads the most onto the 1st PC,I had wanted to see what the result of factor analysis yields-will it be similar or totally different,but got stumped by this error.Can someone please explain what this error means and how do I remove this??
Also,in the output of factanal there is not such thing as the rotated principal components (like the rotated matrix after doing PCA,pca.rotate$loadings in the above code snippet).So how do we see which variable loads onto which factor the most,or does factanal by default give the rotations??