Multi-dimensional K-Means Algorithm for Student Clustering

Abstract

K-Means is one of the popular methods for generating clusters. It is very well-known and commonly used for its convenience and fastness. The main disadvantage of these criteria is that user should specify the number of cluster in enhance. As a repetitive clustering strategy, a K-Means criterion is very delicate to the preliminary beginning circumstances. In this paper, has been proposed a clustering strategy known as Multi-dimensional K-Means clustering criteria. This algorithm auto generates preliminary k (the preferred variety of cluster) without asking input from the user. It also used a novel strategy of establishing the preliminary centroids. The experiment of the proposed strategy has been conducted using synthetic data, which is taken form LIyod’s K-means experiments. The algorithm is suited for higher education for calculating the student’s CGPA and extracurricular activities with graphs.