The Saber 11 state test is characterized not only by measuring the abilities of students in particular situations, but also by characterizing their socioeconomic environment through multiple-response questionnaires attached to the applications, from which di erent socioeconomic levels are generated from the item response theory. In the present work a new methodology is proposed for the determination of those levels using cluster analysis, generating not only a new number of levels but also a new characterization for each one of them. Additionally, the analysis of the results in the area of mathematics from the year 2000-1 to the year 2019-2 is proposed as an academic exercise through the use of univariate time series.