A prototype generation with same class label proportion method for nearest neighborhood classification

The KNN algorithm has a significant effect on classification prediction in Data Mining. In order to solve the drawbacks for KNN algorithm to reduce the costs of the calculation and increase the accuracy, this paper proposed a prototype generation method with same class label proportion for classification to ensure that each class has at least a prototype to be represented. We compare the average success rate of GA, PSO, DE and proposed method SPDE. The experimental results show that the SPDE has more opportunity to do better than others in those problems.