基于K-means算法的优秀班集体评选方法

Excellent Class Selection Method based on K-means Algorithm

传统优秀班集体的遴选方法是将评价班集体的相关属性值进行简单求和,并取总和较大的前K个班级作为优秀班集体,这种评价体系会导致部分班级因为某个属性值过大而成为优秀班集体,而班级其他方面的表现却并不突出,同时,受到年级的限制,部分评价班集体的属性值为空。针对目前优秀班集体评选方法存在的问题,提出基于K-means算法的优秀班集体评选方法,利用同一类簇中的样本相似度较高,而不同类簇的样本相似度较低的聚类特点,从不同年级的样本中评选出相似度较高的K个优秀班集体。通过采用真实样本数据集进行实验的结果表明:基于K-means算法的优秀班集体评选方法能够从不同年级当中遴选出班级各项指标均衡发展的优秀班集体。

The conventional selection method of excellent classes is to sum the relevant attribute values of the classes under evaluation, and take the top K classes with a larger sum as the excellent classes. This kind of evaluation system will cause the problem that the classes having large attribute values and having no outstanding performance in the other aspects will be rated as excellent,whereas some classes will have an empty attribute value under the restriction of lower grades. In order to solve the existing problems,this paper proposes an excellent class selection method based on K-means algorithm. In other words, K excellent classes with higher similarity are selected from the samples of different grades on the basis of the clustering features that samples in the same cluster have higher similarity and samples from different clusters have lower similarity. The results of experiments conducted using real sample data sets indicate that: the excellent class selection method based on K-means algorithm can select outstanding classes from different grades with balanced development of various indexes.