一种基于协同过滤推荐的大学生学业预警系统

A School Precaution System for College Students Based on Collaborative Filtering Recommendation

针对当前大学生学业预警大都仅仅是基于学分成绩而存在片面性且不太准确的问题,本文通过将学生和课程分别类比为“用户”与“商品”,采用基于Item的协同过滤(Item-based collaborativefiltering,ItemCF)推荐技术实现向学生“推荐”最有可能挂科的课程,协助完成大学生的学业预警。该基于ItemCF的学业预警算法在Spark上实现,相比于FP-growth,基于协同过滤推荐等方法的大学生学业预警系统可以获得50.74%的召回率、25.25%的精确度、33.72的F1-Messure和85.19%的覆盖率。

School precaution of current college students shows problems of being credit scores based,one-sided and not accurate,the paper metaphors students and courses as“users”and“commodities”respectively,realizes“recommendation”of most likely failed courses to student with ItemCF(Item-based collaborative filtering),to help them complete school precaution of college students.Compared with FP growth,school precaution system of college students based on ItemCF with Spark realized algorithm can achieve 50.74%recall rate,25.25%accuracy,33.72 F1-messenger and 85.19%coverage rate.