Impact of Time Granularity on Histories Binary Correlation Analysis

Abstract

Activities taken by students within a Virtual Learning Environment (VLE) can be represented by using binary student histories. Virtual Learning Environments allow educators to track most of the students’ individual activities that can be used to elicit the students social communities. In this work, we analyse the impact of granularity in the social community elicitation. Granularity can be seen as the resolution of the student history vectors where each time slot is directly dependent from this value. Indeed, the higher is the resolution of the students histories the more precise is the representation of their actions within the VLE. When comparing the histories using various similarity measures to elicit the students’ groups, we find the optimal granularity and demonstrate that there is a resolution limit where the similarity measures will not help to distinguish the social communities.