We studied short-term electric load forecasting using crafted hourly features across three datasets and four models—XGBoost, LightGBM, Bi-LSTM, and Random Forest—and found that feature engineering significantly improved accuracy over basic features. We also demonstrated that Polar Bear Optimisation (PBO) outperformed or matched PSO and GA for hyperparameter tuning, achieving lower error metrics and particularly strong improvements for the Bi-LSTM model.
Using crafted features and polar bear optimization algorithm for short-term electric load forecast system
We studied short-term electric load forecasting and showed that crafted features and Polar Bear Optimisation–based hyperparameter tuning significantly improved prediction accuracy across multiple machine learning models.