A new minority kind of sample sampling method based on genetic algorithm and K-means cluster

摘要

In view of the classification favors seriously to the most kinds when we use the traditional sorter to classify the imbalanced data set and the errors of classification of minority kind is big, A new minority kind of sample sampling method based on genetic algorithm and K-means cluster is proposed. First the method clusters and groups the minority kind of sample through K-means algorithm, then gains the new sample in each cluster through the genetic algorithm and the valid confirmation is proceed. Finally, The validity of experimental results is proved through using SVM and KNN sorter.