딥러닝 모형을 활용한 교육 빅데이터 분석 논의

Abstract

Big data is being collected across all sectors of society, and the education sector is no exception. The generalization of big data can lead to educational policy decisions based on information derived from big data and enhance the effectiveness of education. Big data in the education field can be roughly divided into three types according to its depth and utilization: macro-level data that includes student (including parents) level, teacher level, and school level data generated and collected by educational institutions; meso-level data for text such as students’ forum discussions, postings, and essay; and small-scale data (such as clickstream) for recording learning activities. In this study, a part of the data from the Korea Education Employment Panel Wave II was fitted to the deep learning model as an example of the spread of the use of big data in education. Using a neural network model including two hidden layers with 140 variables of various student levels such as learning, leisure, and family life as the input layer, the level of student career maturity was predicted and the prediction performance was compared with the multinomial logistic regression model. Discussions related to education big data analysis using deep learning models are presented along with the analysis results.