基于分群的多主体教学评价改进算法

An Improved Algorithm for Teaching Assessment of Multi-subject Based on Clustering

针对多主体给出的评教分数存在不公平的缺陷,提出一种改进的评价算法。该算法首先将不同主体的评教分数线性变换到同一分数空间,并采用k-平均分群法将学生的评价分数分成不同群体,将最小群体的分数排除不参与教师评价分数的计算,根据剩余群中心和群长度加权计算教师的评教分数。通过该算法计算得到教师的排名和同行评价的排名相同,从而进一步验证了本改进算法的有效性。

To overcome the defect of unfairness teaching assessment of different subjects, an improved assessment algorithm was proposed. Firstly, the teaching assessment scores of different subjects were transformed into same score space with linear transformation method. The scores given by students would be clustered into different groups with K-mean clustering. The scores of mini-mum group would be excluded and then the teaching assessment scores were calculated according to the rest group center and group length weight. The rank order of teachers computed by this proposed algorithm is same as the rank order assessed by teachers, and then the effectiveness of this proposed algorithm is validated.