Classification Algorithm Based on Fuzzy Decision for Incomplete Data
Shu Jianhu · 2015
The incomplete data usually exist in traditional Chinese medicine field.The incompleteness of data degrades the learning quality of classification models.Most previous methods dealing with incomplete data only focus on handling incomplete data in the learning phase.For the incomplete data appearing in the classification phase,most of the current approaches cannot work or perform badly.In this paper a novel classifier is proposed to solve the incomplete data classification problem.First of all,a new decision tree for incomplete data is proposed.And then,in contrast to the conventional Boosting algorithm which uses a deterministic decision method during the iterations without considering the incomplete data in the data set sufficiently,we propose a new Boosting algorithm using fuzzy decisions for every hypothesis at the iterations of the Boosting scheme.It selects the data events from a dataset,the weight update mechanism increases the weights of incorrectly classified examples and decreases the weights of those correctly classified examples.The result is aweighted majority vote of the multiple hypotheses.Finally,the experimental results demonstrate the superiority of the proposed strategies for solving incomplete data problem.