Dynamic Classification Trees for imprecise data

This paper provides a supervised classification tree-based methodology todeal withMultivalued data, specifically predictors measurements can be provided bya functional distribution or an interval of values. Main literature refers to symbolicdata analysis, aiming to extend standard methods such as factorial analysis, clustering,discriminant analysis, etc., to deal with symbolic data tables. One approach is todefine a suitable data pre-processing enabling the application of standard methods.A more correct approach is to define suitable methods to deal specifically with unstandarddata. In the framework of supervised classification, there are no proposalin literature for supervised classification methods to deal with both standard andmultivalued data as well. There are only proposals based on data pre-processing.This paper provides a methodology to grow the so-called Dynamic CLASSificationTREE (D-CLASSTREE), upon suitable definition of both a specific splittingcriterion and a tree-growing algorithm. A real world case study will be consideredto show the advantages of the final output and main issues of the interpretation. Acomparative study with older proposals will be also described such to demonstratethe stability and the better accuracy of the D-CLASSTREE.