The generation of energy from renewable sources is continuously growing in Brazil, and consequently the investiments as well. The great quantity of renewable sources, aligned with a campaing for a carbon free energy system show how necessary this type of investment is. Beyond hydroelectricity, responsible for most part of the energy generation in Brazil, in the last decade both Eolic and Solar generation gained much space in the Brazilian Energetic Matrix. However, while the former shows a lot of variation through the days/years, the latter has a characteristic behaviour, where there is no generation during the nigth. For a better understanding of the characteristics of both sources (wind speed and solar irradiance, respectively), this study aims at identifying similar days, that can depict the main behaviour of each source. The data analysed represents the wind speed in fout sites where big Wind Farms ate installed in Brazil, there in the Northeast Region and one inthe South Region, and the irradiance in four sites where big Solar Farms are installed in Brazil, three in the Nothast Region and one in the Southeast Region, all from 1980 to 2020. Machine Learning methods with unsuoervised learning, that aggregate data with similar behaviours, were used. Two of them, K-Means and K-Medoids, are classified as Partitioning Methods, the Ward, that is a Hierarchial Method, and one Model-Based Method (Self-Organizing Maps) were used. It can be seen that, for most methods, the wind speed can be represented by only two days, and the irradiance between two and three days. Therefore, future studies, such as planning, designing, operation and assessmet of renewable-based energy system, can be simplified by using only the representative days instead of the whole dataset, decreasing computation effort and time.