What question did this study set out to answer?

The aim is to develop a framework that predicts students' academic performance using advanced machine learning techniques.

March 12, 2026

A High‐Performance Intelligent Framework for Predicting Students' Academic Performance Using Binarized Simplicial Convolutional Neural Networks in Educational Data Mining

Key Points

The aim is to develop a framework that predicts students' academic performance using advanced machine learning techniques.
Gathered data from the xAPI-Educational Mining Dataset
Pre-processed data using Time Frequency Domain Polarization Filtering (TDPF)
Utilized Weighted Leader Search Algorithm (WLSA) for feature selection
Applied Binarized Simplicial Convolutional Neural Network (BSCNN) for predictions
Optimized model parameters with Triangulation Topology Aggregation Optimizer (TTAO)
Achieved 22.46%, 28.42%, and 25.27% higher accuracy than existing methods
Predicted academic performance levels classified as Low, Medium, and High
Analyzed performance metrics including accuracy, recall, F1 score, precision, and Kappa Value

Abstract

ABSTRACT Educational data mining (EDM) plays an important role in analyzing academic data to enhance teaching strategies, enhance learning experiences, and support institutional decision‐making. Traditional approaches often rely heavily on demographic and academic records, which may not fully capture the factors influencing student outcomes. To overcome these issues, this manuscript proposes a High‐Performance Intelligent Framework for Predicting Students' Academic Performance Using Binarized Simplicial Convolutional Neural Networks in Educational Data Mining (PSAP‐BSCNN‐EDM). At first, the input data is gathered from the xAPI‐Educational Mining Dataset. The data is then pre‐processed utilizing Time Frequency Domain Polarization Filtering (TDPF) to clean and enhance its quality. After pre‐processing, the data is passed through the Weighted Leader Search Algorithm (WLSA) to select the most relevant features from xAPI‐Educational Mining Dataset. The selected features are subsequently fed into the Binarized Simplicial Convolutional Neural Network (BSCNN) to predict students' academic performance levels classified as Low, Medium and High. Generally, BSCNN does not express any adaptation of optimization strategies for figuring out the best parameters to ensure precise academic performance prediction. Hence, Triangulation Topology Aggregation Optimizer (TTAO) employed to optimize the weight parameter of BSCNN. The proposed PSAP‐BSCNN‐EDM method is implemented and the performance metrics such as Accuracy, recall, F1 score, precision, and Kappa Value are analyzed. Then the performance of the proposed PSAP‐BSCNN‐EDM method attains 22.46%, 28.42% and 25.27% higher accuracy when compared with existing approaches.

Demander à l'IA

Bookmark