基于分布式数据流聚类的成绩层次化评估方法  

Hierarchical Evaluation Method for Grades Based on Distributed Data Stream Clustering

作　　者：陈海英 陈华 CHEN Haiying;CHEN Hua(Public Course Department,Tianmen Vocational College,Tianmen,Hubei 431700,China;School of Mathematical and Physical Sciences,Wuhan Textile University,Wuhan,Hubei 430200,China)

机构地区：[1]天门职业学院公共课部,湖北天门431700 [2]武汉纺织大学数理科学学院,湖北武汉430200

出　　处：《保定学院学报》2024年第4期105-110,共6页Journal of Baoding University

基　　金：湖北省高等教育学会教育科研课题立项“数字化转型背景下高职教师信息化教学能力提升路径研究”(2023XD120)。

摘　　要：为提高成绩评估结果的准确率,针对成绩数据具有的多元化、容量大、分布式的特点,引入分布式数据流聚类算法,优化数据聚类精度和时效性,实现成绩层次化评价.首先,通过聚类特征指数直方图建立分布式数据流挖掘矩阵,提高成绩数据的聚类性能.将一阶线性微分方程结合分数阶累加,构建全局聚簇模型,保证聚类性能稳定.然后,采用层次分析法处理成绩数据,将成绩按照基础课程、专业课程及实训课程分类,确保层次化评估的精确率,再结合权重法得到成绩评估结果.实验结果表明,该方法有效实现对成绩数据的层次化评估,精确率、召回率、AUC值分别可以达到98.75%、86.67%、0.987,评估时长仅为488.4 ms,且该方法的收敛效果较优,数据流能耗较低.In order to improve the accuracy of performance evaluation results and address the diverse,large capacity,and distributed characteristics of performance data,a distributed data flow clustering algorithm is introduced to optimize the accuracy and timeliness of data clustering and achieve hierarchical evaluation of performance.Firstly,a distributed data flow mining matrix is established through clustering feature index histograms to improve the clustering performance of grade data.By combining first-order linear differential equations with fractional order accumulation,a global clustering model is constructed to ensure stable clustering performance.Then,the analytic hierarchy process is used to process the score data,dividing the scores into three parts:basic courses,professional courses,and practical training courses to ensure the accuracy of the hierarchical evaluation.Then,the weight method is combined to obtain the score evaluation results.The experimental results show that this method effectively achieves hierarchical evaluation of performance data,with accuracy,recall,and AUC values reaching 98.75%,86.67%,and 0.987,respectively.The evaluation time is only 488.4 ms,and the convergence effect of this method is excellent,with low data flow energy consumption.

关 键 词：分布式数据流 聚类算法 成绩评估 直方图 挖掘矩阵 层次分析法 全局聚簇模型 

分 类 号：O29[理学—应用数学] TP399[理学—数学]