A hybrid data mining for predicting scholarship recipient students by combining K-means and C4.5 methods

Abstract

This scholarly investigation delves into the strong desire for academic scholarships within the student body, especially prominent among socioeconomically disadvantaged individuals. The study aims to formulate a hybrid data mining paradigm by synergizing the K-means and C4.5 methodologies. K-means is applied for clusterization, while C4.5 facilitates prediction and decision tree instantiation. The research unfolds in sequential phases, commencing with data input and progressing through meticulous pre-processing, encompassing data selection, cleaning, and transformation. The novelty lies in successfully integrating the K-means and C4.5 methodologies, culminating in the hybrid data mining method. The dataset comprises 200 students seeking scholarships, revealing effective stratification into three clusters—cluster 0, cluster 1, and cluster 2—with 119, 48, and 33 students, respectively. The K-means method proves highly suitable, especially when combined with C4.5, for predicting scholarship recipients. A subset of 81 students from clusters 1 and 2 undergoes predictive modeling using C4.5, resulting in a commendable 85% accuracy, with 17 accurate forecasts and 3 minor inaccuracies. This research significantly enhances scholarship selection efficiency, particularly benefiting socioeconomically disadvantaged students.