A preliminary study of different data mining techniques applied to the case of strong temporal dependence, large number of dimensions and irregular sampling is presented. In particular, data corresponding to the characterization of the Earth's ionosphere have been used.