Losing more than 20 percent of its students each semester, the student attrition in private graduation courses challenges its institutions management. Different approaches to address this problem have been used. To retention marketing management the identification of students is the first necessary step to apply a personalized interaction strategy. In this sense, this work uses a quantitative methodology to classify its students by risk of attrition. Based in historic data of former students of an institution, models were generated by machine learning algorithms and its results compared. Then they were used to classify active students in the educational institution. Afterwards, their lifetime value were estimated in order to help in the definition of retention strategies.