zbMATH — the first resource for mathematics

Problems of clustering and recent advances. (English) Zbl 0571.62051
New trends in data analysis and applications, Adv. Ser. Manage. 2, 167-182 (1983).
[For the entire collection see Zbl 0564.00022.]
We briefly go over the three following points:
1) What are the basic choices that the user who wishes to classify multidimensional data will have to make at the beginning?
2) We then show that most problems arising in applications can be formulated in terms of the optimization of a mathematically definable criterion.
3) Last we show that in many statistics problems (probabilistic or not) we can obtain an additional understanding if we take the clustering point of view, which emphasizes local properties which turn out to be important in practical problems.
62H30 Classification and discrimination; cluster analysis (statistical aspects)