A Preliminary Study on Clustering Student Learning Data

Clustering techniques have been used on educational data to find groups of students who demonstrate similar learning patterns. Many educational data are relatively small in the sense that they contain less than a thousand student records. At the same time, each student may participate in dozens of activities, and this means that these datasets are high dimensional. Finding meaningful clusters from these datasets challenges traditional clustering algorithms. In this paper, we show a variety of ways to cluster student grade sheets using various clustering and subspace clustering algorithms. Our preliminary results suggest that each algorithm has its own strength and weakness, and can be used to find clusters of different properties. We also show that subspace clustering is well suited to identify meaningful patterns embedded in such data sets.