Search

Educational data mining is valuable for uncovering latent relationships in educational settings, particularly for predicting students' academic performance. This study introduces an interpretable hybrid model, optimised through Tree-structured Parzen Estimation (TPE) and Support Vector Regression (SVR), to predict overall scores (OT) utilising five assignments and one examination mark as predictors. Neural Network-based, Tree-Based, Ensemble-Based, and Boosting-based methods are evaluated against the hybrid TPE-optimised SVR model for forecasting final examination grades among 492 students enrolled in the TPP7155 (General Science) course at the University of Southern Queensland, Australia, during the 2020-2021 academic year. Additionally, Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive explanations (SHAP) techniques are employed to elucidate the inner workings of these prediction models. The findings highlight the superior performance of the proposed model, exhibiting the lowest Root Mean Squared Error (RMSE) and Relative Root Mean Squared Error (RRMSE), as well as the highest Willmott's index (WI), Legates–McCabe index (LM), and Nash–Sutcliffe Efficiency (NS). With assignment and examination marks identified as pivotal predictors of OT. SHAP and LIME analyses reveal the examination score (ET) as the most influential feature, impacting predicted OT by an average of ±4.93. Conversely, Assignment 1 emerges as the least informative feature, contributing merely ±0.64 to OT predictions. This research underscores the efficacy of the proposed interpretable hybrid TPE-optimised SVR model in discerning relationships among continuous learning variables, thereby empowering educators with early intervention capabilities and enhancing their ability to anticipate student performance prior to course completion.

This paper develops a trustworthy deep learning model that considers electricity demand (G) and local climate conditions. The model utilises Multi-Head Self-Attention Transformer (TNET) to capture critical information from G, to attain reliable predictions with local climate (rainfall, radiation, humidity, evaporation, and maximum and minimum temperatures) data from Energex substations in Queensland, Australia. The TNET model is then evaluated with deep learning models (Long-Short Term Memory LSTM, Bidirectional LSTM BILSTM, Gated Recurrent Unit GRU, Convolutional Neural Networks CNN, and Deep Neural Network DNN) based on robust model assessment metrics. The Kernel Density Estimation method is used to generate the prediction interval (PI) of electricity demand forecasts and derive probability metrics and results to show the developed TNET model is accurate for all the substations. The study concludes that the proposed TNET model is a reliable electricity demand predictive tool that has high accuracy and low predictive errors and could be employed as a stratagem by demand modellers and energy policy-makers who wish to incorporate climatic factors into electricity demand patterns and develop national energy market insights and analysis systems.