STUDENT DROP OUT ANALYSIS

Abstract

Student dropout remains a persistent challenge in educational systems, affecting institutional performance and student career outcomes. This study proposes a datadriven framework for early prediction of student dropout using a hybrid Machine Learning and Deep Learning approach, with primary reliance on the Random Forest algorithm. The proposed mechanism integrates academic indicators, attendance patterns, and socio-economic attributes to construct a comprehensive student profile. Data preprocessing techniques, including normalization, missing value handling, and categorical encoding, are applied to ensure data consistency and reliability. The Random Forest classifier is employed due to its robustness in handling high-dimensional data and its ability to minimize overfitting through ensemble learning. Additionally, feature importance analysis is utilized to identify critical factors influencing dropout risk. To enhance analytical depth, a comparative evaluation with a baseline deep learning model is incorporated, enabling performance benchmarking across different learning paradigms. Experimental results demonstrate that the proposed Random Forest-based model achieves superior predictive accuracy and stability, effectively identifying at-risk students at an early stage. The system provides actionable insights for educators and administrators to implement timely intervention strategies. This research contributes to the development of intelligent educational analytics systems aimed at improving student retention and academic success.