Influential factors in the desertion of electronic engineering students from UPTC admitted in 2015

This work identifies the impact that the entry score, the origin place, the application option and the socioeconomic level have on the academic desertion of the electronic engineering students of the UPTC, admitted during 2015. The KNN (K Nearest Neighbors) algorithm was implemented to classify students as deserters and no-deserters. The selected factors for the study made it possible to establish a prediction of deserter and no-deserter students with a prediction error of 4.17%. The factor that most influences the prediction of student desertion was the admission score. A prediction error of 12.5% was obtained when independently relating this factor, which indicates that the program admission score is the most relevant indicator to predict the desertion. The factor that has the least relationship with desertion is the student's origin place with a prediction error greater than 40%. In this study, the most significant causes of student desertion in the program were identified in order to propose follow-up and accompaniment strategies to students and thus reduce desertion rates.