Study On Student Behavior Analysis And Prediction Based On WiFi Data

When dealing with a large number of archival data information,colleges and universities usually choose big data technology to help them manage archival information.The school’s wireless network is fully covered.Teachers and students at school can use the wireless network to access the Internet anytime and anywhere,collecting wireless Internet information,and analyzing student behavior.Manage students in colleges and universities to create a good learning atmosphere.This paper conducts cluster analysis and research on the behavior of students using WiFi on campus,and proposes a student learning enthusiasm model based on the GRU-Attention neural network algorithm.The research work is as follows:1.Carry out cluster analysis and research on the behavior of students using WiFi.First,according to the mobile device using the certified WiFi signal,the sico system is used to collect the spatiotemporal information of daily activities generated by students connecting to the network through the wireless AP.Then,remove incomplete data and noise,and analyze the data for sparsity.Finally,use matrix calculation to generate eigenvectors of students’ s patio-temporal information,dig out the distribution of students’ interests from the trajectories of students’ WiFi behaviors,use K-means algorithm to cluster the eigenvectors of student behaviors,and calculate the different types of students on campus every day The proportion of the stay time of the location,the students with different behavior characteristics are obtained,and the students are divided into three types: learning type,closed type and active type.Based on this clustering result,it focuses on the analysis of the differences in the three behavior patterns between time,education,and gender.The cluster analysis results show that the distribution of student behavior data in time,space,and categories has certain rules,reflecting the There is a correlation between the behavior of students using WiFi and their motivation to learn.2.In order to better analyze the correlation between students’ Wi-Fi behavior and learning enthusiasm,the problem of students’ learning enthusiasm prediction is abstracted into a classification problem,and the collected Wi-Fi behavior data is clustered and analyzed,and the student enthusiasm based on GRU-Attention is proposed.Forecast model.First of all,when collecting student behavior data,according to the students’ use of campus WiFi in the learning place,extract the characteristics of the student’s stopping time in the campus learning place,the frequency and purpose of surfing the Internet,etc.,and thus the degree of concentration of the student’s learning.Then,perform data cleaning and sparsity analysis on the collected student behavior data,use the K-means algorithm to preprocess the feature data set,divide the students into three positive types,A,B,and C,and introduce student performance as a reference standard to form A dataset of student spatiotemporal behavior characteristics.Finally,the data set is input to the GRU information extraction layer,and then after the weighting process of the attention layer,the probability distribution of the classification is calculated by softmax.This paper conducts experiments on a self-built student behavior data set,and obtains the distribution rules of the behavior data of students using WiFi in time,space and category through the cluster analysis method,integrates the attention mechanism in the GRU,and uses the GRU-Attention network model.Student learning motivation prediction.The experimental results show that the accuracy of the GRU-Attention neural network model is better than other common prediction models.The student’s use of WiFi can predict their learning enthusiasm,which helps the instructor guide the student to establish a good style of study.