2.6.1. PCA usage within Uranie

Let’s use a relatively simple example to illustrate the principle and the way we can achieve a reduction of dimension while keeping the inertia as large as possible. One can have a look at a sample of marks from different pupils, in various kinds of subject, all gathered in the Notes.dat whose content is shown below.

#TITLE: Marks of my pupils
#COLUMN_NAMES: Pupil | Maths | Physics | French | Latin | Music
#COLUMN_TYPES: S|D|D|D|D|D

Jean 6 6 5 5.5 8 
Aline 8 8 8 8 9
Annie 6 7 11 9.5 11
Monique 14.5 14.5 15.5 15 8
Didier 14 14 12 12 10
Andre 11 10 5.5 7 13
Pierre 5.5 7 14 11.5 10
Brigitte 13 12.5 8.5 9.5 12
Evelyne 9 9.5 12.5 12 18 
          

One can have a look at some of these variables against one another to have a sense of what’s about to be done. In Figure 2.54, the maths marks of the pupils are displayed against latin on the left and physics on the right. Whereas no specific trend is shown on the left part, an obvious correlation can be seen from the right figure meaning one can try extrapolate the maths marks from the value of the physics one.

../../_images/PCA_notes_examples.png

Figure 2.54 Representation of some variables of the Notes sample.