Educational Data Mining using Correlation based Feature Selection and classification for Future Learning Prediction

Abstract

Educational Data Mining (EDM) is a growing field focused on analysing and modelling educational data to improve teaching and learning processes. This study explores the application of Correlation-based Feature Selection (CFS) and Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) for future learning prediction in educational systems. The research aims to predict student performance by leveraging historical data and identifying the most relevant features through correlation analysis. The CFS technique plays a crucial role in reducing the dimensionality of the dataset by selecting features that are highly correlated with the target variable while eliminating irrelevant or redundant ones. This improves the quality of the data fed into the predictive model and enhances the model's accuracy. The proposed model uses RNN-LSTM, a powerful deep learning architecture known for its ability to capture long-term dependencies in sequential data. LSTM, a special type of RNN, is particularly suitable for educational data that often involves time series, such as student performance over multiple periods. By training the model with student data, it can effectively predict future learning outcomes, providing valuable insights into individual student progress. The results demonstrate that the integration of CFS and RNN-LSTM achieves an impressive prediction accuracy of 97.10%, indicating the effectiveness of this approach for educational data mining tasks. The model’s high accuracy signifies its potential in providing actionable insights for educators to tailor learning strategies and interventions. This study highlights the significance of feature selection techniques and deep learning models in educational data mining, offering a promising solution for enhancing educational outcomes and personalized learning experiences.