2014HS: 33065 Applications of Artificial Intelligence

Hello,

sorry for being away, but the questions were asked quite late ;)

I really like the answers and discussion. Thank you for that. For PCA, I would like to try another way as well (the ellipsoid is a bit misleading, as we do not really enclose the data): Actually, you could also say (as in the second part), you look where the highest variability of data is located and make the first axis in this direction. This axis conveys the most information about the data and is thus assumed to be most useful for classification.

Best regards,

     Marcus