Application of the improved K-means clustering algorithm in support vector machine