The role of superficial geology of the subsoil in studies of seismic microzonation is of great interest because it is a latent threat before an earthquake. With non-invasive techniques, the behavior of seismic waves can be inferred by studying the ways they propagate. The Nakamura technique (H/V) is the most used in order to identify the natural frequency of the site in the seismic microzonation studies. This technique allows to know the surface geology through experimental in-site data and implementing computational techniques. In this paper, records of environmental seismic noise were used for the extraction of characteristics of each of the signal components. The H/V curves allowed to identify the peaks in the frequency range from 0,1 to 10 Hz. A clustering technique has been used on the feature vector to identify different subsets related to similar seismic behavior for the characterization of the subsoil.