On Clustering Algorithms for Cognitive Diagnosis Based on Item Responses

Cognitive diagnosis in education and psychology reveals students’ knowledge grasp, aiding in personalized learning and targeted interventions. As it evolves, researchers are increasingly focusing on nonparametric methods like clustering, which traditionally groups subjects into unlabeled clusters based on similarity. The novel proposed clustering average distance algorithm(C-ADA) is used to label the estimated clustering results based on the subjects’ observed response patterns. At the same time, the corresponding proof is given. It serves as a supportive tool for future clustering in this field. Simulation studies show that the C-ADA resolves the unlabeled problem. The effectiveness and applicability of this method are also investigated.