Comparison of statistical clustering techniques for correction analysis of achievements of the college entrance examination

Abstract

In this study, three different statistical clustering methods: a hierarchical, k-means and an artificial neural network (Self-Organizing Maps, SOM) technique were applied to analyze the achievements of the college entrance examination. A comparison of the methods for correction analysis was attempted. The research results indicate that distance function has an important effect on the correction among the courses, and that the performance of SOM was similar to the cluster result obtained by other two clustering algorithms. The empirical results also show that there is a better correction between Chinese, Science basis and Selective course. © 2011 Springer-Verlag Berlin Heidelberg.