Abstract

A CLUSTERING-BASED ANALYTICAL FRAMEWORK FOR EVALUATING STUDENT ACADEMIC PERFORMANCE

Dr. Siddamsetty Upendra*, S. Mahendra, Dr. K. Murali

ABSTRACT

The digital revolution in education has opened up considerable new prospects for determining academic success through sophisticated data analysis.This study provides a useful data-driven technique for forecasting academic outcomes by grouping comparable learning activities. We used unsupervised machine learning techniques such as K-means, Expectation-Maximization (EM), Canopy, and Hierarchical clustering to detect long-standing patterns in student data that are typically neglected. The study demonstrates how these clustering tools may identify students based on their unique academic profiles utilizing the WEKA analytics platform. Using these ideas, educators can move beyond "one-size-fits-all" approaches and promote more tailored academic planning and timely support. Finally, the results demonstrate that K-means is the most successful technique for institutional use, offering the optimal balance of accuracy and speed.

Keywords: K-means, EM algorithm, clustering algorithms, student performance analysis, and educational data mining.