Clustering method for forming student groups based on the cognitive performance measured using eye responses

In our past study, Eye and Pupil responses were analyzed to measure the ability of the students to learn the meaning of a novel word or concept from the given contextual information in the passage. Moreover, in this paper, the eye response data of several subjects is used to form student groups of varying cognitive levels. Student clusters are formed using clustering methods such as K-Means, K-Medoids, and variants of hierarchical clustering. Moreover, the purity of the clusters is measured using the Silhouette Score (SC), Davies-Bouldin Index (DBI), and Calinski-Harabasz Index (CHI). When two or three clusters are formed using various types of feature sets, the hierarchical clustering method forms good-quality clusters in comparison to K-Means, and K-Medoids. Hence, when student groups are formed based on their cognitive, language, and learning skills, it helps to provide real-time assistive, supplementary learning content to the students on a need basis and may assist in student-team formation based on their cognitive abilities.