This study explores the application of data mining techniques in physics education research, focusing on the analysis of kinematics graph interpretation skills. The research had two main objectives: to demonstrate the utility of RapidMiner, a data mining tool, in analyzing educational data, and to compare its effectiveness with traditional item analysis methods. Fifty-nine Grade-10 students at Chiang Mai University Demonstration School completed the Test of Understanding Graphs in Kinematics (TUG-K) before and after participating in a problem-based learning module integrating high-speed video analysis. Traditional statistical analysis revealed significant improvement in student performance (p<0.001, effect size 0.76). Association rule mining, conducted using RapidMiner, uncovered key relationships between test items that were not apparent through traditional analysis. These relationships provided insights into common student misconceptions and areas requiring targeted instruction. The study demonstrates the potential of advanced data mining techniques to reveal deeper patterns in educational data compared to conventional item analysis methods. This novel application of RapidMiner in physics education research offers a promising approach for more detailed analysis of student understanding, potentially informing more effective teaching strategies and curriculum design in physics education.
Read full abstract