Decision Tree Construction: A continues label support Degree based Approach

Abstract

Data mining and classification systems utilize decision tree algorithms since they proffer rapid speediness, advanced exactness and also simple organization of those algorithms. An ideal decision can be built only when the appropriate attributes are chosen. This paper focuses on throwing light on choosing characteristics based on the theory of attribute support degree on account of which a unique decision tree construction algorithm is proposed on the basis of rough set and granular computing theory. It is henceforth proved that the decision tree proposed by the new approach yields far more better results in terms of precision and consistency as compared to the decision trees yielded by ID3, C4.5 and DTBAS.