Survey of Decision-Tree Method for Structured Continuous-Label Classification

Structured continuous-label classification is a variety of classification in which the label is continuous in the data, but the goal is to classify data into classes that are a set of predefined ranges and can be organised in a hierarchy. In traditional decision (classification) tree algorithms, categorical (class) variable as the label is assumed. When the label is a continuous variable in the data, two achievable approaches based on existing DT algorithms can be used to handle the situations. For the first approach uses a data discretisation method in the preprocessing stage to convert the continuous label into a class label defined by a finite set of non-overlapping intervals and then applies a DT algorithm. In the second approach using the continuous label directly, simply applies a regression tree algorithm. These approaches have their own drawbacks. In the hierarchy, the ranges at the lower levels are more specific and inherently more difficult to predict, whereas the ranges at the upper levels are less specific and inherently easier to predict. Therefore, both prediction specificity and prediction accuracy must be considered when building a DT from this kind of data. In this survey paper, we survey the classification techniques related to the continuous label. More classification approaches are there in the literature. The main goal of the survey is present an overview of previous classification techniques. Various methods that are written by the different authors are presented in this survey.