Decision trees (DTs) have been well recognized as a very powerful and attractive classification tool, mainly because they produce interpretable and well-organized results. In developing DT algorithms, it is commonly assumed that the label (target variable) is nominal or a Boolean variable. In many practical situations, however, there are more complex classification scenarios, where the labels to be predicted are not just nominal variable, but have distance or relation between each other. Since previous studies paid little attentions on this problem, they cannot be used to construct a DT from data with labels of distance concept. To remedy this research gap, this study aims to develop an innovative DT algorithm called “Construct a DT from data with labels of distance concept.” An empirical study was performed to evaluate the proposed algorithm on three real datasets. The experiments show that the proposed method can significantly increase the classification precision without sacrificing the classification accuracy. It is also demonstrated that the classification results can be effectively used for recommendation purposes.
展开▼