Despite the significant benefits of PV systems, they come with challenges. One major issue is their intermittent nature, which complicates matching energy supply with demand and integrating them into electrical grids. Consequently, there has been a notable increase in research efforts aimed at improving solar PV energy forecasting. This study aims to forecast PV Energy using support vector regression (SVR) and extreme gradient boosting (XGBoost), leveraging historical weather data recorded at half-hour intervals and the yield of a grid-connected PV system with amorphous silicon panels. Various metrics, including root mean squared error, mean absolute error, max error, and R-squared, are employed to evaluate and compare the accuracy of these machine learning models. The hyperparameter optimization for both models have been done using grid search. The obtained results indicate that SVR and XGBoost provide exceptional accuracy, achieving a correlation coefficient of 0.99 and 0.999 respectively, in terms of accuracy XGBoost has showed better results in both testing and training phases. Suggesting that they are effective tools for optimizing energy production and enhancing overall solar system efficiency.