Classification is widely used technique in the data mining domain, wherescalability and efficiency are the immediate problems in classificationalgorithms for large databases. We suggest improvements to the existing C4.5decision tree algorithm. In this paper attribute oriented induction (AOI) andrelevance analysis are incorporated with concept hierarchys knowledge andHeightBalancePriority algorithm for construction of decision tree along withMulti level mining. The assignment of priorities to attributes is done byevaluating information entropy, at different levels of abstraction for buildingdecision tree using HeightBalancePriority algorithm. Modified DMQL queries areused to understand and explore the shortcomings of the decision trees generatedby C4.5 classifier for education dataset and the results are compared with theproposed approach.
展开▼
