Francisco da Conceição Silva;Andre Macedo Santana;Rodrigo Miranda Feitosa
{"title":"基于可解释人工智能的中等技术教育辍学指标研究","authors":"Francisco da Conceição Silva;Andre Macedo Santana;Rodrigo Miranda Feitosa","doi":"10.1109/RITA.2025.3566095","DOIUrl":null,"url":null,"abstract":"With the increasing application of Artificial Intelligence (AI) in education, it becomes essential to understand the factors influencing school dropout, as the educational context demands reliable decision-making. This study investigates dropout indicators in secondary-level technical courses at a Brazilian institution, using Explainable AI (XAI) techniques applied to Machine Learning models. The study analyzed data from 15,084 students to identify the main factors contributing to school dropout, utilizing predictive models and applying the explainability techniques LIME and SHAP to highlight key dropout factors, thereby improving prediction transparency. The results show that decision tree-based models performed best, with Random Forest achieving 85% Recall, effectively identifying students at risk of dropout. LIME and SHAP consistently highlighted school attendance, family income, and place of residence as key dropout factors. The analysis of the explainers showed that students with low attendance and lower income are more likely to drop out. These findings highlight the importance of targeted educational policies, such as scholarship programs, transportation assistance, and personalized academic support, especially for vulnerable students. This study contributes to the understanding of factors associated with school dropout and provides insights for the formulation of more effective educational policies. Strategies such as scholarship programs, transportation assistance, and academic monitoring can be implemented to reduce dropout rates.","PeriodicalId":38963,"journal":{"name":"Revista Iberoamericana de Tecnologias del Aprendizaje","volume":"20 ","pages":"105-114"},"PeriodicalIF":1.0000,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10981760","citationCount":"0","resultStr":"{\"title\":\"An Investigation Into Dropout Indicators in Secondary Technical Education Using Explainable Artificial Intelligence\",\"authors\":\"Francisco da Conceição Silva;Andre Macedo Santana;Rodrigo Miranda Feitosa\",\"doi\":\"10.1109/RITA.2025.3566095\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"With the increasing application of Artificial Intelligence (AI) in education, it becomes essential to understand the factors influencing school dropout, as the educational context demands reliable decision-making. This study investigates dropout indicators in secondary-level technical courses at a Brazilian institution, using Explainable AI (XAI) techniques applied to Machine Learning models. The study analyzed data from 15,084 students to identify the main factors contributing to school dropout, utilizing predictive models and applying the explainability techniques LIME and SHAP to highlight key dropout factors, thereby improving prediction transparency. The results show that decision tree-based models performed best, with Random Forest achieving 85% Recall, effectively identifying students at risk of dropout. LIME and SHAP consistently highlighted school attendance, family income, and place of residence as key dropout factors. The analysis of the explainers showed that students with low attendance and lower income are more likely to drop out. These findings highlight the importance of targeted educational policies, such as scholarship programs, transportation assistance, and personalized academic support, especially for vulnerable students. This study contributes to the understanding of factors associated with school dropout and provides insights for the formulation of more effective educational policies. Strategies such as scholarship programs, transportation assistance, and academic monitoring can be implemented to reduce dropout rates.\",\"PeriodicalId\":38963,\"journal\":{\"name\":\"Revista Iberoamericana de Tecnologias del Aprendizaje\",\"volume\":\"20 \",\"pages\":\"105-114\"},\"PeriodicalIF\":1.0000,\"publicationDate\":\"2025-03-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10981760\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Revista Iberoamericana de Tecnologias del Aprendizaje\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/10981760/\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Revista Iberoamericana de Tecnologias del Aprendizaje","FirstCategoryId":"1085","ListUrlMain":"https://ieeexplore.ieee.org/document/10981760/","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS","Score":null,"Total":0}
An Investigation Into Dropout Indicators in Secondary Technical Education Using Explainable Artificial Intelligence
With the increasing application of Artificial Intelligence (AI) in education, it becomes essential to understand the factors influencing school dropout, as the educational context demands reliable decision-making. This study investigates dropout indicators in secondary-level technical courses at a Brazilian institution, using Explainable AI (XAI) techniques applied to Machine Learning models. The study analyzed data from 15,084 students to identify the main factors contributing to school dropout, utilizing predictive models and applying the explainability techniques LIME and SHAP to highlight key dropout factors, thereby improving prediction transparency. The results show that decision tree-based models performed best, with Random Forest achieving 85% Recall, effectively identifying students at risk of dropout. LIME and SHAP consistently highlighted school attendance, family income, and place of residence as key dropout factors. The analysis of the explainers showed that students with low attendance and lower income are more likely to drop out. These findings highlight the importance of targeted educational policies, such as scholarship programs, transportation assistance, and personalized academic support, especially for vulnerable students. This study contributes to the understanding of factors associated with school dropout and provides insights for the formulation of more effective educational policies. Strategies such as scholarship programs, transportation assistance, and academic monitoring can be implemented to reduce dropout rates.