Multivariate Regression Predictive Modeling in Analyzing Student Performance: A Data Mining Approach

摘要

This study is based on student performance in the measured in terms of their academic grades. The performance is measured based on a number of demographic, behavioral and historical attributes. To analyze the result and to create a predictive model, the study uses the data mining techniqueof multivariate regression and correlation attribute evaluation on a dataset of 395 respondents drawn from an academic program covering the subject of mathematics in a school in Portugal. A total of eight predictor variables are used to predict the criterion variable. The criterion or thedependent variable is final grades. Out the eight attributes, five attributes i.e., “student’s age,” “quality of family relationship,” “going out with friends” and “current health status” are significant predictors of “final grades”whereas three attributes i.e., “free time after school,” “work day alcohol consumption,” “weekend alcohol consumption” and “number of school absences” are insignificant as far as the regression model is concerned.