School dropout is a global problem that affects all countries, being worst in underdeveloped countries like those in Africa and Latin America mainly, also in some countries in Asia. This paper presents a methodology based on multiple criteria decision making (MCDM) and machine learning approaches which address the problem of assessing factors related to student attrition in universities for aiding in developing efficient policies to mitigate dropout rates. As a real case study data from the "Universidad Simon Bolivar" was developed to demonstrate the application of the proposed methodology. Some interesting findings of factors related to student dropout are discussed.