Predict Student Achievement Through The Campus E-card Data

With the gradual and in-depth development of modern campus construction,the use of the one-card campus has infiltrated into major universities.The data of college students' life and learning traces are stored and recorded,and a large amount of information data is generated.These data hide many important characteristics related to the achievements,which are often ignored in the past,but the student scores are not only related to past achievements,It is also deeply related to student learning and life behavior.With the development of data mining and machine learning technology,we can use the massive data recorded by the card to explore the behavior characteristics of students,predict student performance through student behavior,and provide the basis for teaching management and daily management and decision-making in colleges and universities.As a basic criterion for judging a student,academic achievement is also the main reference for college scholarship evaluation.Predicting student achievement in advance can effectively give early warnings for students' academic work,urge students to study and improve their professional level.The learning style and living habits can reflect the students' achievements to some extent.This paper uses data mining and machine learning to research on the impact on student performance by E-card data,such as library entry records,early,middle and late dinner times,early departure time,book borrowing record,classroom study times,and combined with the results of the first and second semester to predict the students' third grade results.Educational data mining is an important application direction in data mining,It is the use of statistics,machine learning and data mining methods on learning behavior and learning process to quantify,analysis and modeling.At present,the data mining mainly focuses on the analyzes of grades or the analysis of previous achievements.Based on the previous research,this paper explores the characteristics of students' learning behaviors,using data miming algorithms Decision trees ?random forests to predict the student score with historical achievements.The results of the study showed that the top 20% and the last 20% of the students in the rankings had a significant difference in the number of times they entered the library,the number of meals on time in the morning,at the middle and the evening,and the number of books borrowed.The top 20% of the classmates prefer to enter the library,borrow more books,and work habits are more punctual.Therefore,student achievement is not only related to past achievements,but also closely related to students' learning behaviors and living habits.