Analysis of characteristics influencing student dropout in the context of a latin american university

Análisis de características que influyen en la deserción estudiantil en el contexto de una universidad latinoamericana

The present work aims to deepen the study of the student dropout, which is a serious problem that worries the governments, university institutions and students worldwide. To achieve the above, this study uses data mining to analyze student dropout in a Latin American university by discovering the most influential relevant characteristics and by identifying patterns to facilitate the understanding of such problem. The methodology used considers an adaptation of the steps proposed by KDD (knowledge discovery in databases) and the study design was observational, descriptive and cross-sectional, using convenience sampling. The sample is made up of 10705 students, which are distributed in 7 faculties and 33 undergraduate academic programs. A model based on decisióntree was used to verify the predictive relationships between the status of dropping out student and the influential characteristics. As a result, this work identified techniques and methods commonly used in these studies and developed a method to identify patterns of relationships among the most influential characteristics in the student dropout. We found that the main influencing characteristics in this study refer to socioeconomic level, gender, employment status and academic performance. One aspect to be highlighted is the coincidence of the findings of this study with the results of other similar studies worldwide, in which academic performance was identified as a fundamental factor that affects university dropout. As a conclusion we can state that university student dropout is caused by a set of characteristics and their interrelations rather than a single characteristic.