Solar Photovoltaic has been used for long due to potential shortage of fossil fuel energy, its effect on the environment, and the increase in energy consumption around the world. Solar PV power output is irregular in nature due to the intermittency and the other dependent factors, like wind velocity, irradiance, ambient temperature, humidity, etc. Due to uncertainty, it is challenging to predict power output from solar photovoltaic. The electric power system in Rwanda is facing a challenge of power outages that is caused by several reasons including the intermittence of energy from solar PV power plants which result to load shedding in different regions of the country. To attain a sustainable solution to this problem; prediction of the power output from solar PV can be utilized which may help the utilities to plan the scheduling of the power system around the country in a predictive approach. This can be achieved by intelligent algorithm which is used to analyze the data of the recent five years. For both metrics (i.e., R2-SCORE and RMSE) used to evaluate the three machine learning methods, the KNNR with parameter K=8 is the best model. It achieved R_SCORE of 0.978 as shown in Table 2.  K in this case means that it uses 8 data points that are nearby the input of the concern. These algorithms provide the prediction model which can be considered as formula which is very useful to the utilities in decision making.

Key words: Machine learning, solar photovoltaic, renewable energy, metrics, regression model