Research On The Method Of Determining Poor Scholarship Based On Data Mining

With the continuous expansion of the higher education,a special group is also expanding-poor students.Poor students become a social problem that drawing the attention of people around the nation.Education fairness is the basis of social fairness.In order to promote the fairness of education,the state,universities and the community work together to research and establish a relatively perfect student subsidy policy system,which ensures the smooth enrollment of students and the successful completion of their studies.However,there are still shortcomings in the identification of poor students and the granting of grants:the grants fail to cover all the poor and the non-poverty students are subsidized by the state.The community continue to offer advice for the precise funding,aimed at achieving the goal of national precision funding.In this paper,using the statistical methods and data mining theory,combined with a university card data and poor grants to data.(1)using the box-line rough outline of the school in the precise financing of poor students,indicating that there is a perfect space for the university to help the poverty;(2)through the one-way analysis of variance and contingency table analysis,theoretically proved that students consumption level and students.Academic achievement is one of the factors that affect the identified poor students,that the poor students not only to support the family financial difficulties of students living difficulties also have to finance the ability to pay their development costs of the development of difficult students;(3)combined with K-means clustering students in accordance with the consumption of their consumption patterns of clustering,compared with the lowest consumption of the population,the accounting for 25.07%of the total population,more in line with the proportion of poor students in China,but the proportion of funding in fact did not meet this.Then building poor students forecast Model to further help schools,the community and the government to do a better job of subsidizing poor students to ensure the work of poor students to complete their studies.In this paper,three classification models are constructed:logistic regression model,Naive Bayesian algorithm and K-nearest neighbor algorithm.The recall rate,accuracy and F1 value of the three models are obtained.By comparison,we find that the K-nearest neighbor algorithm can judge better Whether the students are poor students,the accuracy rate of 75.28%,check the rate of 85.35%.