Towards a Sustainable Science Education Laboratory for the SDGs: Photovoltaic Energy Sufficiency and Machine Learning Forecasting

Laboratories are a crucial element in higher education as centers for science learning, research, and innovation. However, laboratories are known to consume approximately 60–65% of a university’s total electrical energy. Therefore, the transition towards Sustainable Science Education Laboratory (SEL) is urgent; one promising way is the utilisation of renewable energy photovoltaic (PV) systems. To evaluate the energy sufficiency of PV, accurate predictions of the fluctuating electrical energy consumption patterns of SEL are required. This study conducted a short-term forecast of electrical energy demand using the Extreme Gradient Boosting (XGBoost) machine learning algorithm. By applying rolling statistics during feature engineering stage, despite the relatively small dataset the model succesfully produced accurate predictions, achieving a Mean Absolute Error (MAE) of 78.91, a Mean Absolute Percentage Error (MAPE) of 6.06%, and an overall accuracy of 93.94%. These results demonstrate that the proposed method is effective for estimating short-term energy demand. This study supported to the achievement of SDG 7 and SDG 4, especially on higher education. Moreover, the forecasting model and Python program developed in this study can serve as a ground for researchers and academics to assess the sustainability readiness of their laboratories. Nevertheless, this study is limited to the context of electricity. Future studies should broaden their scope to consider including other sustainability indicators, such as the carbon footprint, waste management, and academic activities.

Перевод пока недоступен