Research On University Students' Employment Prediction Model And Application Based On Decision Tree Algorithm

The implementation of the policy of enrollment expansion in higher education has led to a significant increase in the number of college graduates year after year,but the employment situation of college students is not optimistic.Although the advancement of education informatization has led to the accumulation of a large number of student information in college employment guidance systems,including various school-based data and related employment data,in actual applications,these data have not been fully utilized.Data mining technology,as a new information processing technology formed by the cross of multiple disciplines,can effectively mine the value behind the data.Applying this technology to employment work in colleges and excavating the implicit knowledge and rules behind student information data,not only can enrich the application of data mining related theories,but also can provide powerful decision support for employment guidance in universities,and have important theories and practical significance.Based on the idea of data mining,this paper takes the relevant data of the master graduates of the XXXX College of Educational Technology in the past three years as the research object,and uses the classic C4.5 algorithm,constructs the college students"whether can be employed successfully" and the "employment area" decision tree prediction model.The specific research contents are:(1)Data preparation for college graduate employment forecasting model and correlation analysis of each influencing factor.Through the literature analysis method,the relevant factors that affect the employment of college students are combed,based on this to determine the source of the database.In order to form a standard data set,a series of pre-processing processes,such as collection,integration,conversion,and clean-up of the information data are executed,and using correlation analysis method to quantify each influencing factor,and selecting the important test attributes that can be used to construct the predict model.(2)The construction of university students' employment forecasting model.Using the classic decision tree C4.5 algorithm,and taking the normative data sample as the mining object,construct two decision tree prediction models of "whether can be employed successfully" and the "employment area",and extract the classification rules according to the employment prediction model.(3)Evaluation and application analysis of college students' employment prediction model.The forecasting accuracy and influence factors of the prediction model were analyzed.Based on the prediction model,the relevant employment situation of 2018 graduates was analyzed,and provided relevant decisions and suggestions for the employment of college graduates.The forecasting accuracy of the two university graduate employment prediction models are all above 85%,which is in line with the expected design.Through the factor analysis of the prediction model,the following are the important factors affecting whether or not the college students can be successfully employed:"achievements" and "cadre conditions".The important factors affecting the selection of employment areas for university students are "source of origin information" and "position matching".The innovation of the paper are as follows:This paper proposes the use of correlation analysis method for attribute screening,which ensures the scientificity of the prediction model,develops prediction tools to improve the efficiency of the construction of the prediction model,all in all,the study organically combines relevant theoretical techniques of data mining with employment issues of university students,and providing an application model that can be used for reference.