Optimizing the proportion of prototypes generation in nearest neighbor classification

Most of the methods for prototype generation that gives a suggestion for the proportional to classes label is equal to the average, but does not completely arrive at ideal accuracy. In this paper, we modify the encoded form of the individual to combine with the proportion for each class label as the extra attributes in each individual solution, besides the use of the DE algorithm with the Pittsburgh´s encoding method that include the attributes of all of the prototypes and get the perfect accuracy, and then to raise up the rate of prediction accuracy. The second contribution of this paper is find out that for each numeric attribute value should be normalized to transform to the range [¿1, 1] that get the better accuracy result than the range [0, 1].