Hernández Londoño, Jesús D.

Profile Picture

Filters

20221
Reset filters

Settings

Citations

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Desarrollando modelos de análisis predictivo para mejorar la probabilidad de graduación de estudiantes subgraduados
    (2022-11-14) Hernández Londoño, Jesús D.; Rivera Santiago, Roberto; College of Arts and Sciences - Sciences; Rolke, Wolfgang A.; Lorenzo González, Edgardo; Department of Mathematics; Parés, Carmen P.
    University graduation rates in Puerto Rico are alarmingly low. On average, around 45 % of undergraduate students obtain their degree at 150 % program length [43]. This percentage becomes even more dramatic when viewed in conjunction with the school dropout rate. The 40 % of students don’t finish high school and the remaining 60 % do finish high school, but not all go to college. Suppose that from the 60 % everyone gets into college. So of that 60 % just 45 % graduates at 150 % of the duration of their academic program. Although some colleges offer workshops and mentoring to students to help them graduate, there is no objective way to predict whether a particular student will complete her degree. For such reasons, this research project aims to implement predictive models using machine learning methods to predict whether a student will complete their university degree at UPRM 150 % program length. These models were applied to 24,432 data of students admitted to the Mayaguez University Campus (RUM) between 1999 and 2010; which includes variables such as the student’s study program, standardized entrance test scores, parent’s level of education, among others. 6 machine learning methods including Gradient Boosting and TabNet (a new artificial neural network architecture for tabular data) were used to predict whether an undergraduate student graduates at 150 % program length from UPRM. Gradient Boosting showed the best performance of all the methods when predicting whether a student graduates at 150 % program length from UPRM. Said tool allows university officials to detect specific students and develop intervention strategies to increase their chances of graduation.