Variables que influyen en el rendimiento de los estudiantes de postgrado: Una perspectiva desde la analítica del aprendizaje

In the last decade, the use of learning analytics and the management of large volumes of data have contributed substantially to the way higher education institutions track, analyze information and predict student performance (Clow, 2013). The objective of this work was to identify the variables that influence the academic performance of graduate students, through the application of learning analytics techniques (Chatti, et al., 2012). The algorithms selection of attributes and decision trees (Witten, Frank, Hall, Pal, 2016) were applied to a sample of data collected from 136 graduate students in a simple random way. It was identified that in general they prefer to study in the afternoon and that they invest 43.83% of their time in the review of the course content while they are active in the platform; 10.92% of the time they participate in forums and 31.10% of the time they carry out activities. Through the algorithm of attribute selection, the four most important variables that influence performance are defined, namely: total time invested in the course of course consultation, elaboration of tasks, participation in forums and teamwork. Also, applying decision trees, 6 patterns are established that determine some final note, whose most important variable is the total time spent on the platform. Finally, it is determined that the variables: time invested in the platform in the consultation of content, teamwork, tasks and forum activity, positively influence the satisfactory performance of the graduate student and those variables related to the consultations, time and day of study do not intervene in such performance, these findings give the guideline to focus efforts on building meaningful content and tasks focused on achieving the desired learning supported by team activities.   Keywords: learning analytics; academic performance; usage patterns; postgraduate students.