The energy sector is currently changing towards renewable energy, such as photovoltaic solar, to the detriment of traditional fossil fuels. This growth is driven by the need to reduce greenhouse gas emissions. However, solar energy production has an inherent dependence on weather conditions. Indeed, increasing the ambient temperature beyond a certain threshold can lead to a significant reduction in the efficiency of photovoltaic cells. This phenomenon is linked to the increase in the electrical resistance of the semiconductor materials constituting the cells, which limits the circulation of electrical charges, thus compromising the production of electricity and reducing the efficiency of photovoltaic installations. The aim of this work is to contribute to the search for a model for predicting maximum powers in photovoltaic systems, using methods based on approaches offered in machine learning. This manuscript presents an analysis of the state functioning in photovoltaic cells under the effect of temperature’s variations, linked to meteorological censuses over the last 45 years in six regions of North-West Algeria. A parametric study is carried out using the PVsyst software to indicate the significant impact of different factors on energy production. Next, an assessment of the prediction models is developed using machine learning. At the end of this study, two prediction models are compared and tested in order to satisfy the concordance and reliability of the results.