Comprehensive Evaluation Of Poverty Students In Colleges And Universities Based On Logistic Regression

According to the survey,the number of needy students in ordinary colleges and universi-ties is relatively large,therefore,how to subsidize poverty students has become a very im-portant issue.The assessment of impoverished students is the premise and difficulty of the funding for poverty students in universities.At present,there are many disadvantages in many universities,such as subjective basis and objective basis.Nowadays,the use of big data min-ing technology to evaluate poverty students has been an important application of big data technology in campus.The campus card consumption data of our real example analysis,from data prepro-cessing,data import,algorithm design,model establishment to the impoverished mining algo-rithm to achieve the whole process.First,to build a big data analysis platform by Tencent will cloud server,campus card data after preprocessing,using Sqoop tool to import large data analysis platform and stored in HDFS.Data mining is carried out based on Spark MLlib ma-chine learning library,and the mining results are displayed.In this paper,we first use the Logistic regression algorithm in the machine learning li-brary to train,and then introduce the SVM algorithm into the traditional Logistic regression model,the new integrated discriminant analysis rule is used to mine poverty students.By cal-culating the consumption amount of students' canteens,the proportion of supermarket con-sumption and the cost of campus network,a feature vector is formed.Based on these eigen-vectors,data sets are constructed to excavate poverty students.The model was established through experiments,and finally the new consumption data were used to predict whether the data correspond to students who are poverty students,and the results are explained and ana-lyzed.