A Partial Coverage Based Approach to Classification
Yu Huang, Gongde Guo, Daniel C. Neagu · 2007
The k-nearest neighbour (kNN) method is simple but effective for classification. The bottleneck of kNN is it needs a good similarity measure which could be problematic in some cases especially for datasets containing categorical data. In this paper, a partial coverage based classificaiton (PCC) method is proposed which works without similarity measure and conversion for categorical data. Moreover, the PCC method is easy to be implemented. Experiments were carried out on some public datasets collected from the UCI machine learning repository. The experimental results show that the proposed method is better than some classical classificaiton algorithms in terms of classification accuracy. The PCC is a quite promising method for classification.