Novel Decision Tree Classifier for Data with Distance Labels
Hao Hu, Chia‐Chi Wu · International Journal of Digital Content Technology and its Applications · 2012
Decision trees (DTs) are powerful classification tools capable of returning well-organized, interpretable results. When developing DT algorithms, it is commonly assumed that the label (target variable) is a nominal or Boolean variable. In many practical situations, however, more complex classification scenarios exist, in which the labels are not just nominal variables, but have distance or shared relationships among one another. To compensate for a lack of research on this issue, this study developed an innovative DT algorithm which is use to Construct a DT from data with labels of distance concept.” We performed empirical analysis on three real datasets to evaluate the proposed algorithm. Results demonstrate that the proposed algorithm can significantly increase the precision of classification without sacrificing accuracy, and the results can be use for the purpose of recommendation.