SAPPNet: students' academic performance prediction during COVID-19 using neural network.

Area of Science:

• Educational Technology
• Artificial Intelligence
• Data Science

Background:

• Predicting student academic performance is challenging due to various factors.
• Existing predictive models often fail to meet educational management needs.
• The COVID-19 pandemic has introduced new variables affecting student engagement and outcomes.

Purpose of the Study:

• To propose a novel deep learning (DL) model, Students Academic Performance Prediction Network (SAPPNet), for accurate grade prediction.
• To evaluate SAPPNet's performance against classical machine learning (ML) and other DL models.
• To leverage a comprehensive dataset including demographic, digital tool usage, and psychological factors.

Main Methods:

• Developed SAPPNet, a DL model incorporating spatial convolution modules for static features and temporal modules for dynamic changes.
• Utilized the Jordan University dataset, encompassing pre- and post-COVID-19 information on student attributes and behaviors.
• Compared SAPPNet with ML models (SVM, KNN, Decision Tree, Random Forest) and other DL models (ANN, CNN, LSTM).

Main Results:

• SAPPNet demonstrated superior performance compared to all benchmarked methods.
• The model achieved high accuracy, precision, recall, and F1-score.
• The integration of spatial and temporal modules significantly enhanced prediction capabilities.

Conclusions:

• SAPPNet offers a significant advancement in predicting student academic performance.
• The model's ability to capture spatial and temporal dependencies provides valuable insights for educational management.
• This research opens new avenues for analyzing educational datasets and improving student support systems.