Decision Tree Model for Educational Services Predicting Children’s Academic Performance by Income Group

Abstract

Background and objective This study used the decision tree analysis among data mining techniques to determine whether children’s academic performance can be classified and predicted by income group based on factors of educational services. Methods For empirical analysis, data from the 10th Panel Study on Korean Children collected in 2017 was utilized. A F test was conducted to analyze the differences in variables by income group, and a decision tree analysis was conducted on the cost and time of private education services utilized by children to predict their academic performance by income group. Results First, as a result of analyzing the research variables by income group, there was a significant difference in institute time, community center time, institute cost, lesson cost, after-school cost, and culture center cost. Second, as a result of the decision tree analysis that predicts children’s academic performance by income group, it was found that for children in the low-income group, institute cost, institute time, visiting cost, and after-school time were important variables that predict their academic performance. For children in the middle-income group, institute cost, after-school time, and after-school cost were important variables for predicting academic performance. For children in the high-income group, the important variables were institute time, institute cost, after-school time, and after-school cost. Conclusion There was no significant difference in children’s academic performance in the earlier grades of elementary school, but there was a significant difference in the private education service they utilized, which may affect future income gaps as well as education gaps. This suggested the need to diversify and improve the quality of public education services as a countermeasure for the fact that parental income will cause an academic gap among children through private education.