A hierarchical shrinking decision tree for imbalanced datasets
Chien-I Lee, Cheng‐Jung Tsai, Chiu-Ting Chen · 2006
Abstract:- Since the real-world datasets are often predominately composed of majority examples with only a small percentage of minority/interesting examples, data mining researchers have put more and more attention on developing efficient approaches to handle the imbalanced datasets. In this paper, we proposed Hierarchical Shrinking decision tree algorithm, called Hshrink, to solve the class imbalance problem. HShrink hierarchically groups minority examples together by using the splitting function derived from geometric mean in each internal node of the decision tree. Consequently, HShrink can accurately mine the rules of minority examples and reach a higher predicted accurately. Key-Words:-, data mining, classification, decision tree, imbalance