Early Warning Technology For Student Dropout Prevention In South West China

Dropout rate is one of the most important indicators that Bureau of Education in Southwest China concerns. Therefore, how to prevent dropout is a vital question need to be answered. Recent years, educational management information systems have been established in many areas, so it is possible to find factors which related to dropout rate, and then prevent it from happening by early warning. This early warning technology can help educational decision makers make more effective polices.This paper is a report of early warning technology research for preventing students dropping out of school. It first discusses the necessity and feasibility of the research, and the problem need to be solved, from the perspective of educational practice. Then, it describes how to use business intelligence technology along with data mining algorithms to analyze the massive student data to find data patterns of dropout. Finally,based on the patterns, it provides case study of how to do the early warning. It also discusses various issues that need to be considered.The dropout early warning technology proposed here is embodied in four stages: Stage one, establishing the relational table of data mining models, building mining models based on the data fields of education management information system, and cleaning up the data of education management information system for the source data, populating the newly created data mining model relational table; Stage two, using decision trees algorithm, Bayesian algorithm, association rules algorithm, clustering algorithm and neural network algorithm to construct data mining models, and correspondingly interpreting and analyzing the results of the five mining models; Stage three, after mining accuracy validation of these five data mining models, it is discovered that mining model constructed by Bayesian algorithm outstands other algorithms for producing more accurate results; Finally, use the Bayesian model and data to carry out dropout early warning. The outcome of this research can enhance educational leader's decision-making capacity, and improve the quality of education management.