Scholar Dropout at General Mathematics subject: identification of relevant variables for its prediction
محفوظ في:
| المؤلفون: | , |
|---|---|
| التنسيق: | artículo original |
| الحالة: | Versión publicada |
| تاريخ النشر: | 2025 |
| الوصف: | The aim of this study was to determine the most important variables for predicting student dropout from the General Mathematics course (MAT001) of the Universidad Nacional de Costa Rica (UNA), considering the types of students and the time at which dropout takes place. Six predictive models were constructed (two student groups at three different times) and three supervised learning algorithms were implemented in each model: Logistic Regression (LR), Random Forest (RF) and XGBoost (XGB). The total sample was split into training files containing data on students who enrolled in the course during the years 2017 and 2018, and test files with data corresponding to students who enrolled in the year 2019. Once the hyperparameters were fitted (10-fold validation), the main variables associated with student dropout (SD) in the General Mathematics course of each model were identified based on the Gini importance measure; performance of the algorithms ranged from F1-Scores of 0.6251 to 0.7300. In addition, the predictive power of the algorithms in each model were compared by means of a repeated-measures ANOVA with 10-fold cross-validation, and no significant differences were found between the three algorithms in any of the proposed models. The main variables associated with student dropout (SD) were academic, such as grades on the academic attitude test (AAT), high school education grades, and grades on MAT001 tests, student attributes as sex and age at enrollment, economic factors such as scholarships and the Social Development Index (SDI), and institutional factors such as high school educational opportunities that students were exposed to, and the ages and specializations of the teaching staff. Based on the results of this analysis, it is recommended that teachers specialized in Educational Mathematics be assigned to teach initial courses, and to propose designs for decision making about actions that increase permanence. |
| البلد: | Portal de Revistas UCR |
| المؤسسة: | Universidad de Costa Rica |
| Repositorio: | Portal de Revistas UCR |
| اللغة: | Español Inglés |
| OAI Identifier: | oai:portal.ucr.ac.cr:article/61275 |
| الوصول للمادة أونلاين: | https://revistas.ucr.ac.cr/index.php/educacion/article/view/61275 |
| كلمة مفتاحية: | Higher Education Student Dropout Student Performance Statistical Data Statistical Methodology Mathematics Enseñanza superior Abandono escolar Rendimiento escolar Datos estadísticos Metodología estadística Matemáticas |