PCA is a $[$ Select $]["$supervised$",$ "unsupervised"$]$ algorithm. The attributes calculated via plain PCA are $[$ Select $]["$nonlinear$",$ "linear"$]$ transformations of the original data.$$ The new attributes are $[$ Select $]["$reductions$",$ "projections"$]$ of the original attributes in the direction of maximum $[$ Select $]["$average$",$ "variance"$]$ of the original data.$$