The student desertion in undergraduate programs of most institutions of higher education in
both Colombia and Latin America is a problem that has a multidimensional impact on the social
and economic development of a country. Despite this, few studies have been conducted at the
Nariño University regarding this problem to implement effective strategies to help minimize this
phenomenon and lead to improving the quality of education at the university.Through data mining
techniques it is possible to predict when a student is going to abandon their studies, applying the data
stored in the databases of an educational institution, to take early action to enable it to reduce this factor. This paper describes the knowledge discovery process that is held at the Nariño University in determining the university community profiles low academic performance and student desertion using the historical database of undergraduate students. This process is supported by TariyKDD, a data mining tool for free distribution, developed in the KDD laboratories of Research Group GRIAS of the Department of Systems of Engineering Faculty.