This study focuses on the methodology for developing a model to forecast electricity production from photovoltaic (PV) panels through the analysis and processing of meteorological and historical data. An important aspect is the application of the XGBoost algorithm, with a detailed discussion of its selection and the process of hyperparameter tuning. The study also presents an approach to integrating various data sources, including retrieving meteorological data from the Open-Meteo API and combining it with data on PV energy production. The resulting model demonstrates high predictive performance, confirming the effectiveness of XGBoost in capturing complex, non-linear relationships even with limited or noisy training data.