Analyzing and Comparing Clustering Algorithms for Student Academic Data

The online learning data could be analyzed in order to improve the teaching-learning process. The Performance of students depends on several factors, that could be managed to improve their academic growth. It is difficult to acquire a complete picture of the state of the student's performance and concurrently uncover critical information from their patchy performance when using the typical student grouping based on average scores. Clustering algorithms can be used to identify patterns and clusters in learners’ data in e-learning scenarios. There are several clustering algorithms that can be used to do this task. In this paper, k-means clustering, hierarchical clustering, and the farthest first clustering algorithm will be used to analyze students’ academic data and form clusters based on 4 major categories: demographic features, academic background, parent participation, and behavioral features. The dataset used in this study is taken from Kaggle, which is originally collected from LMS (Learning Management system) Kalboard-360 and it consists of 480 student records with 16 features. The main aim of this study is to identify the clusters with given features and the best-performing clustering algorithm which gives the highest correctly classified instances. Results show that the highest correctly clustered instances are given by the k-mean clustering algorithm.