Learning Process Analysis Based on Sequential Pattern Mining and Lag Sequential Analysis in a Web-Based Inquiry Science Environment

Abstract

To explore what learning behaviors affect learners science-based inquiry learning performance, this work develops an xAPI (Experience Application Programming Interface)-based learning record store module embedded in a Collaborative Web-based Inquiry Science Environment (CWISE) to record students detailed learning process data, including learning behaviors, learning time, and learning sequence. More importantly, based on the recorded learning behaviors, this work discusses the significant correlation and cause-effect relation among science inquiry ability, learning time, and learning performance, and to explore the remarkable shift and difference in the learning behaviors of learners with different learning performance and inquiry abilities by using sequential pattern mining and lag sequential analysis. The research results show that inquiry ability, total learning time, and inquiry simulation experiment learning time present notably positive correlations with learning performance, with linear regression, showing the predictability of learning performance, on which inquiry simulation experiment learning time appears the most significant effect. Moreover, based on sequential pattern mining, learners with different learning performance do not show significant differences on the learning sequence of the entire inquiry learning course, but follow the inquiry learning process sequence in the curriculum design. Also, based on lag sequential analysis, learners with high learning performance and high inquiry ability would re-adjust the previous hypotheses after proceeding the simulation experiment of buoyancy, while learners with low learning performance and low inquiry ability are lack of such a key inquiry learning behavior.