Combining decision trees classifiers: a case study of automatic sleep stage scoring

Abstract

This paper presents a new approach of classification in which multiple decision trees are combined together for achieving better accuracy compared to that achieved by each of the individual constituent decision trees. A major unit of the proposed system is the combination unit for which we present two algorithms; one is based on pre-pruning and true positive rate and the other is based on maximum probability voting. In presenting this new method, we use the case study of sleep stage scoring as a basis of demonstration. For such a task, two tree classifications are utilised. We performed a tree classification based on the training data and then combined the resulting model with another classification tree supplemented by the expert according to Rechtschaffen and Kale’s sleep scoring rules. We applied this method to nine recordings, six of which were used to construct the training tree and the remaining three were used for testing. The experiments showed that the combination method has a 7% better accuracy over a single model.