Clustering social media user for grouping students in final project using K-Means Clustering and Support Vector Machine

Abstract

As social media has become inseparable attribute from a person, it is used in common to identify someone’s behavior. This practice can be seen in Human Resource Development upon recruiting new employee or when an emigration officer checking someone in the airport or applying visa. This paper propose model to group students based on their preference of twitter account. The steps are by collecting dataset from news website and use each category to label the data. After dataset is collected, user classified according to the friend list and then cluster the users based on the classification result. Test shows that SVM has better accuracy compared to other algorithm, while the elbow method to determine number of cluster does not show best k number since graphic of the method shows exponential form. For future works, it is recommended to use silhouette method to determine k form clustering and Latent Dirichlet Allocation (LDA) to label dataset into multi-label data.