Application of Cluster Analysis to Distance Education Students

Abstract

Educational databases often have hidden knowledge about the students, their academic behavior, their study skills and their performance in an academic programme. If this knowledge could be made explicit, it might be useful to the educational institution for facilitating learning. There are many studies that have tried to analyse students’ characteristics to draw conclusions about the students. These studies were mainly based on either nominal or interval data and the characteristics were judged by the percentage of students possessing these characteristics. A negligible number of studies have tried to analyse students globally with respect to all their characteristics. The question thus arises as to whether students can be classified on the basis of pre-existing knowledge contained in the institution’s student database. One of the data mining techniques of structuring data is “cluster analysis”. Clustering literally means “gathering” or “drawing together”. In terms of data, clustering means dividing the data in such a way that similar data points come together. The present study reviewed data on various student characteristics and attempted to apply cluster analysis on the science graduate students of Indira Gandhi National Open University, India from the Delhi region. The dataset involved a sample of 75 students from the data repository of 1,307 students of which 693 students were active students. The study explored various clusters of students that were “heterogeneous among” them but “homogenous within” themselves. The preliminary results obtained are an attempt to identify the similarities among the students, which bind them together in a particular cluster, despite their being different in their performances academically. This study has also tried to build up a measure to indicate the dissimilarity among the students.