Power Utilities Energy Data Acquisition and Forecasting: GUI Development and ML Evaluation

Electric load forecasting is critical for power system planning and operation, especially given the rising electricity consumption and integration of renewable resources. Machine learning (ML) is widely used for energy forecasting, but the lack of consistent and complete historical data can hinder ML model research, as fragmented and flawed datasets may degrade model performance. This study presents a graphical user interface (GUI) to access reliable historical load data from power utilities, aiding in ML model development. The developed GUI was tested for retrieving data from two utilities—New York Independent System Operator (NYISO) and Hydro-Québec (HQ). Four ML models—linear regression, random forest, eXtreme gradient boosting, and long short-term memory—were used to evaluate data quality and forecasting performance. Results show high prediction accuracy, with mean absolute percentage error (MAPE) typically below 5% and coefficients of determination (R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup>) close to 1. This study underscores the GUI's effectiveness in simplifying data collection while highlighting the importance of data availability and quality in achieving accurate results.