Research on the Financial Difficulty Level Recognition of Needy Students Based Upon the Decision Tree C5.0

Abstract

Accurate recognition on the needy students is a core part of the college student subsidy and management work. Supported by the predictive capability of data mining model, this thesis studied the data sample of the recognition on needy students in a college. It selected 33 explanatory variables and one target variable through data pretreatment. Then by means of decision tree C5.0, the data sample was used to build the model. By calculation, the predictive accuracy of decision tree C5.0 was close to 90% on identifying the financial difficulty level of needy students. Therefore, it possessed excellent predictive effect. In addition, Kappa coefficient evaluation method was used to further prove that the model possessed favorable predictive effects. This study aimed to provide decision basis for subsidizing the needy students in colleges and universities, consequently improving the “targeted subsidy” work.

A Hybrid Clustering Model for Determining Financially Needy Students for Sponsorship

“…Previous works for determining needy students are based on classification methods such as factorization machines [1], deep learning [2], decision tree C5.0 algorithm [3], random forest and logistic regression [4]. These related works determined needy students from family income, family expenditure and level of borrowing.…”