Forecasting solar photovoltaic power output is a difficult task, but it is essential for effective solar energy planning. Artificial intelligence and machine learning algorithms are promising to improve the forecasting accuracy. In this work, we discuss how different types of data (e.g. weather, solar irradiance, etc.) can be used to improve the accuracy of the forecast. We evaluate the effect of using machine learning algorithms, including artificial neural networks, support vector machines, and gradient boost to design the model that effectively represent the output power of a photovoltaic solar system in Jordan. Experimental results show that algorithms such as gradient boost perform the best compared to simulation and other machine learning algorithms. Best model selection applies 10-fold cross validation using data from weather stations to forecast PV power output. The best model has been trained using imputed and pre-processed data from 2018-2022 and has shown excellent results. The Gradient Boost algorithm outperformed other algorithms, with an RMSE value 0.808 and 0.921 R2. This suggests that the Gradient Boost algorithm is a promising tool for forecasting PV power output.