Decision tree for classification and regression: A state-of-the art review

Abstract

Classification and regression are defined under the umbrella of the prediction task of data mining. Discrete values are predicted using classification techniques, whereas regression techniques are most suitable for predicting continuous values. Analysts from different research areas like data mining, statistics, machine learning, pattern recognition, and big data analytics preferred decision trees over other classifiers as it is simple, effective, efficient, and its performance is competitive with others in a few cases. In this paper, we have extensively reviewed many popularly used state-of-the-art decision tree-based techniques for classification. Additionally, this work also reviews some of the decision tree based techniques for regression. We have presented a review of more than forty years of research that has been emphasized on the application of decision tree in both classification and regression. This review could be a potential resource for all the researchers who are keenly interested to apply the decision tree based classification/regression in their research work.