International Journal of Advanced Engineering, Management and Science (IJAEMS) Peer-Reviewed Journal ISSN: 2454-1311 | Vol-7, Issue-9; Sep, 2021 Journal Home Page: https://ijaems.com/ Article DOI: https://dx.doi.org/10.22161/ijaems.79.1 This article can be downloaded from here: www.ijaems.com 1 ©2021 The Author(s). Published by Infogain Publication. This work is licensed under a Creative Commons Attribution 4.0 License. http://creativecommons.org/licenses/by/4.0/ Clustering of Learners based on Readiness to Online Modality using K-Means Algorithm Daryl B. Valdez, Rey Anthony G. Godmalin 1 BSCS Department, Bohol Island State University – Clarin Campus, Philippines Received: 22 Jul 2021; Received in revised form: 22 Aug 2021; Accepted: 01 Sep 2021; Available online: 08 Sep 2021 Abstract— Clustering is one of the important techniques in data mining. It is an unsupervised task of grouping similar data. It has been applied in various fields with high degree of success. This study aimed to determine the learner segments based on readiness to online learning modality using K-means algorithm. A dataset was collected, tabulated and pre-processed. Further, the values were scaled and transformed using t-distributed Stochastic Neighbor Embedding. Using elbow method and determining the silhouette score, the best K value was determined. Then clustering was conducted using the selected number of clusters. Results revealed three groups of learners; Moderate-signal mobile users, Low-signal mobile users, and mixed group of Low/moderate-signal mobile/broadband users. Students from the different clusters are more suited for flexible learning as opposed to online learning. Varied learning modalities can be catered for students from the different learner segments. Formulation and adoption of new policies are needed to offset the effect of the pandemic towards the students. Keywords— Clustering, K-means algorithm, data mining, online learning modality, learner’s segmentation. I. INTRODUCTION Clustering is an unsupervised task of dividing data points into a fixed number of groups wherein the data points of a group bears close similarity and are different from those in other groups (Syakur et al, 2018). K-means algorithm is one of the methods of clustering data. It is the most commonly used clustering method due to its speed and simplicity (Yuan et al, 2019). Clustering has a variety of applications in various fields including; market segmentation, medical imaging, social network analysis, image segmentation and anomaly detection. Not only that, recent studies revealed that it can also be useful in the field of academe. A study was conducted and used clustering to classify learners according to learning style preferences (Pasina et al, 2019). Results of the study revealed student outliers which have different learning style from the rest allows instructors to properly address their concerns. Further, clusters of students with similar learning styles allows ease of work on class assignments. Another study was also conducted using hierarchical clustering in grouping students according to learning style (Yotaman et al, 2020). The experimental results show that grouping students into seven clusters using the Euclidean distance function and the ward linkage criteria yields the highest efficiency in clustering. The resulting clusters can help identify the behaviors and learning skills of students which will enable teachers more options in selecting and using appropriate methods and teaching strategies. Aside from segmenting learners, cluster analysis can also be used in the other aspects of the student-learning environment such as in determining groups of teachers according to some factors. In fact, a study was successfully conducted using clustering to group teachers. Further, the results were used as basis for evaluating teaching quality (Sangita et al, 2011). Other studies involve clustering of educational aspects in the case of online learning. Studies were conducted using clustering algorithms in determining user groups and personalized intelligent tutoring. Clustering algorithm was modified by exploiting the use of minimum spanning tree. Results revealed increased performance over traditional clustering algorithms when used in online learning resources (Wu et al, 2016).