This research work presents a comparative analysis in order to evaluate the impact of adding photovoltaic panel temperature as an additional explanatory variable in models that already include solar radiation to predict the electrical power generated by photovoltaic systems. Two experiments were carried out: the first using only solar radiation, and the second incorporating both solar radiation and panel temperature. Regression techniques, such as linear regression, K-nearest neighbours algorithm, decision tree and random forest, were applied to model the power generation. The data, collected by sensors installed in a bioclimatic house at 10-minute intervals over one year, showed that the incorporation of panel temperature improves the accuracy of the model for all the techniques evaluated. In particular, the K-nearest neighbours model, configured with three neighbours and no weighting, showed the best performance, achieving the lowest Mean Squared Error and the highest Coefficient of Determination (R \(^2\) ). Future work will investigate also the inclusion of additional environmental variables and the application of different machine learning techniques to improve performance.