This report focuses on results obtained from a classification technique applied to time series data in a medical ill-structured The statistical analysis and classification --in ill-structured-- of such data are often inadequate because of the intrinsic characteristics of those domains. The database in this analysis contains information relative to patients with major depressive disorders or esquizofrenia; as a consequence, a high quantity of database variables contain data corresponding to measures taken in different instant of time, making curves. For this reason we are motivated about how we can establish a useful classification technique of curves in a medical ill-structured domain.