摘要
针对传统的求总分统计成绩方法的不足,提出了一种基于K-Means算法的成绩聚类分析方法。该方法根据成绩分布情况选取固定的初始聚类中心,改进了K-Means算法随机选取初始聚类中心导致聚类不稳定的不足,在聚类后通过聚类内差异与聚类间差异的比值来衡量聚类的质量。通过一个实例说明了该方法在分析学生成绩数据中的应用,实验结果表明,聚类方法比传统的求总分方法更合理、更科学,聚类结果蕴含更多有用的信息,而且改进后的聚类方法降低了随机选取初始聚类中心所产生的结果的不稳定性,聚类效果较好。
Aiming at the shortage of traditional scores summing method, a K-means based clustering method to analyze scores is proposed. In this method, the initial clustering centers is fixedly selected by considering the distribution of students＇ scores. The clusters＇ instability caused by randomly choosing the initial clustering centers is improved. The ratio of the difference within clusters to the difference between clusters is used to measure the quality of clusters. An example is given to explain the application of the improved algorithm on analyzing students＇ scores. Experiments show that clustering method is more reasonable and scientific when compared with traditional scores summing method, and the clustering results contain more useful information. Experiments show that the improved clustering algorithm reduces the instability of the clustering result from randomly choosing the initial clustering centers.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第5期1130-1133,共4页
Computer Engineering and Design
基金
广西民族大学青年科研基金项目(0509QN32)
