Linking prediction with personality traits: a learning analytics approach

Abstract

Open, flexible and distance learning has become part of mainstream education in China. Using a blended learning program in a Chinese high school as the case, this study adopted data-mining approaches to establish predictive models using personality traits. Results showed that, for students with high OE and low extraversion, and students who are low on both of these constructs, the number of postings in digest (NPD) and average score of after-class test (SCT) were significant predictors of their achievement. For students with low OE and high extraversion, time spent on viewing course resources and number of answers provided in the format of text were significant predictors. For those with high OE and low extraversion, time spent on learning online and number of questions raised in the format of hypermedia, NPD and SCT were significant. Furthermore, deep belief networks performed best in identifying at-risk students at each stage.