An algorithm for generating fuzzy decision tree with trapezoid fuzzy number-value attributes

This paper presented an algorithm based on the L-R-COG using the information entropy minimization heuristic for generating decision tree with the trapezoid fuzzy number-value attributes. We first define the L-R-center of gravity (L-R-COG(Atilde)) of the trapezoid fuzzy number-value attribute Atilde=(a, b, c, d), and next compute its L-R-COG(Atilde), in the end we use the information entropy Enty(Atilde, T, S) minimization heuristic to choose the test attribute for generating decision tree. By considering the L-R-COG(Atilde) and analyzing non-stable points, the presented algorithm gives us a desirable behavior of the information entropy of partitioning. Finally, an example shows the utility of the proposed algorithm. Specially, if d=c=a=b to Atilde=(a, b, c, d), the corresponding learning algorithm is the learning algorithm of decision tree generation for continuous-valued attributes [2].