Optimization and Improvement Based on K-Means Cluster Algorithm

K-means cluster algorithm is one of important cluster analysis methods of data mining, but through the analysis and the experiment to the traditional K-means cluster algorithm, it is discovered that its cluster result varies along with the initial selected cluster central point, and the difference is big. In view of this question, this text proposed the method of seeking the initial cluster center embarking from the data object distribution, moreover in order to accurately appraise the cluster result, it also proposed cluster assessment method based on the data object. Through analyzes and contrast of the experiment, the improved cluster algorithm surpasses the traditional K-means cluster algorithm, and also can obtain high and stable classified accuracy.