Research On The Construction Of College Students' Classification Prediction And Portrait Based On Students' Daily Achievements

With the continuous development of educational informationization,the data of teaching management in Colleges and universities is increasing rapidly.Using scientific methods to get valuable knowledge and information can provide better services for college teachers.Through the daily learning data of students,data mining technology is used to analyze the data and establish relevant models,to explore the classification of students and its influencing factors,and establish curriculum portrait for students,so that teachers can timely and accurately grasp the learning effect of students,and carry out differentiated teaching for students with different degrees of mastery,so as to improve the teaching quality.The main work of this paper is as follows:(1)Collection and processing of student data.Collect the teaching data of students in the course of university computer foundation,select and fill in the abnormal data such as missing value by data preprocessing,and unify the dimension by normalization method for each module score;then determine the main factors affecting the classification of students according to the weight value of each index output by XGBoost model,so as to provide basis for the determination of virtual index in user portrait.(2)The Stacking model is established to predict the classification of college students.Based on the single classifier model,this study proposes a Stacking classification model based on the integration of XGBoost model,Random Forest model and LR model,which can predict the classification of students.The accuracy of Stacking model in the experiment of predicting poor students and excellent students is 85.92%and 83.23%,respectively.Through comparing with other classification models by contrast,it is found that the students' classification model based on Stacking algorithm is more accurate and has better performance than other classification models,which is more suitable for students' classification and prediction experiments.Finally,the accuracy of the model is verified by the data of 2019 students.(3)Establishing students' curriculum portraits.Through the methods of data statistical analysis,data mining technology and visual display,the course portraits of students are constructed.Through the four indicators of basic attributes,original basis,discipline mastery level and learning input level,the basic information,learning habits,discipline advantages and disadvantages,and the degree of effort of students are understood,so as to help teachers understand students comprehensively.Finally,the accuracy and effectiveness of the portrait are verified by comparison.Through the research of this paper,it is found that the stacking classification algorithm based on the course results of university computer foundation can accurately reflect the results of students' classification,and the student user profile established can help teachers to clearly understand the learning characteristics of students,which is of great significance for future teaching research.