Tests for Decision Tables with Many-Valued Decisions - Comparative Study

The paper is devoted to the study of a greedy algorithm for construction of approximate tests (super-reducts). This algorithm is applicable to decision tables with many-valued decisions where each row is labeled with a set of decisions. For a given row, we should find a decision from the set attached to this row. The idea of algorithm is connected with so-called boundary subtables. After constructing a test we use algorithm which tries to remove attributes from a test and obtain a reduct. We present experimental results connected with the cardinality of tests and reducts for randomly generated tables and data sets from UCI Machine Learning Repository which were converted to decision tables with many-valued decisions. To make some comparative study we presents also experimental results for greedy algorithm which constructs a test based on generalized decision approach.