This article is cited in 6 papers

Classification based on full decision trees

I. E. Genrihov^a, E. V. Dyukova<sup>b

a</sup> Moscow State Pedagogical University, Malaya Pirogovskaya ul. 1, Moscow, 119991 Russia
^b Dorodnicyn Computing Center, Russian Academy of Sciences, ul. Vavilova 40, Moscow, 119333 Russia

Abstract: The ideas underlying a series of the authors’ studies dealing with the design of classification algorithms based on full decision trees are further developed. It is shown that the decision tree construction under consideration takes into account all the features satisfying a branching criterion. Full decision trees with an entropy branching criterion are studied as applied to precedent-based pattern recognition problems with real-valued data. Recognition procedures are constructed for solving problems with incomplete data (gaps in the feature descriptions of the objects) in the case when the learning objects are nonuniformly distributed over the classes. The authors’ basic results previously obtained in this area are overviewed.

Key words: precedent-based pattern recognition problem, full decision tree, entropy branching criterion, voting vertex of a decision tree.

UDC: 519.712

Received: 13.04.2011
Revised: 06.09.2011

 English version: Computational Mathematics and Mathematical Physics, 2012, 52:4, 653–663

Bibliographic databases:

• 
• 
• 
• 
• 
•