Statistical Learning and Optimization Methods in Data Mining
G.-W. Weber, P. Taylan, S. Ozogur, B. Akteke-Ozturk
to appear in the book of the Turkish Association of Statiticians at the Occasion of "Graduate Summer School On New Advances in Statistics"
The notion of data mining is one of the modern ciphers of interdisciplinary research which employs the applied mathematical and computational statistics. It deals with the outcomes of experiments, records, measurements, questionnairs, etc., and it aims at modeling and prediction. This paper presents three core elements of mathematical data mining: clustering, classification and regression. Each of these areas is very large, such that we can only discuss selected aspects here. This selection is done by our methods coming from statistical learning and optimization theory, and by our choice of, e.g., SVMs and MARS which also serve for a unified view. In all the three areas, we display and propose advanced future research.