Academic Early Warning Model for Students Based on Big Data Analysis

Abstract

How to identify in advance and help college students with academic difficulties is an important topic for current education departments and universities. Academic early warning system based on big data analysis comprehensively analyzes the learning, life and psychological data of college students, effectively identifies potential academic problems, and helps teachers and student managers take measures in advance to improve the education quality. The existing academic warning models of college students based on big data analysis often have defects, such as data quality issues, lack of key variables, nonlinear problems, and human factors. Therefore, this paper aimed to study the academic early warning model of college students based on big data analysis. After elaborating on the key points of collecting the academic early warning model data based on big data analysis, this paper explained the reasons of calculating the Pearson correlation coefficient of collected big data. This paper constructed an academic early warning model of college students based on deep self-coding network, provided the construction process, and explained its working principle. After optimizing the model parameters, this paper analyzed the model reconstruction error based on sliding window statistical method, and further improved the prediction ability and generalization performance of evaluating the deep self-coding network model, thus obtaining higher academic early warning accuracy. The experimental results verified that the constructed model was effective.