This article presents a novel approach to forecasting energy production using machine learning techniques with feature selection. The study utilizes machine learning algorithms optimized using historical weather and real-time energy data. Feature selection is performed through a mathematical framework based on correlation coefficients and mutual information, ensuring the use of only the most significant predictors. SHAP (SHapley Additive exPlanations) is employed to analyze the impact of selected features on prediction outcomes, enhancing model transparency. Experimental results show high forecasting accuracy and robustness, highlighting the role of meteorological variables in energy production. This work advances predictive modeling for renewable energy systems and provides a comprehensive framework for identifying key factors influencing PV energy production, supporting improved forecasting and decision-making in energy management.