Discarding Variables in a Principal Component Analysis. Ii: Real Data

摘要: In this paper it is shown for four sets of real data, all published examples principal component analysis, that the number variables used can be greatly reduced with little effect on results obtained. Five methods discarding variables, which have previously been successfully tested artificial data (Jolliffe, 1972), are used. The compared and to satisfactory real, as well artificial, although none overwhelmingly superior others.