Predição de aprovação em um curso em tecnologia da Informação no Instituto Metrópole Digital da UFRN: uma aplicação da análise de regressão logística

• Autor(a) principal: Carelli, Diego Silva Campos
• Data de Publicação: 2017
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da UFRN
• Texto Completo: https://repositorio.ufrn.br/handle/123456789/34318
• topic: Analise de resíduos; Curva ROC; Diagnósticos; Educação a distância

A common school performance indicator in educational learning processes is the rate of approved students. Relating this rate to student's characteristics can provide a means to understand the approval process and to improve student's achievement. In this work, logistic regression analysis is applied to model the probability of approval of 1.270 students in a technical course of the Digital Metropolis Institute of the Federal University of Rio Grande do Norte. The logit of the probability of approval is estimated by a linear combination involving the main effects of the variables gender, age, number of course absences, argument in the institution admission exam, duration of the high school course, instruction of the student's father, as well as the interactions sex and age, sex and number of absences and age and number of absences. The suitability of the model fit is confirmed by the Hosmer-Lemeshow statistics, examination of the model residuals and analysis of diagnostic measures. Also, the model has specificity of 84.3%, sensitivity of 94.2%, overall rate of correct classifications of 89.2% and area under the ROC curve equal to 0.956, suggesting therefore the model has high predictive ability. In short, keeping the effects of the other variables fixed, it is observed that the chance of approval estimated by the model: (i) increases with the increasing of age among male students; (ii) decreases with the increase in the number of absences, especially among female students; (iii) increases with the increase of the argument of entrance in the institution; (iv) is lower among those who do not attend high school or attend high school but have not graduated yet than the chance of approval of students who have taken three years or less to complete high school; (v) is lower among those whose father's education is complete high school, do not know or the father is deceased than the chance among students whose parents are illiterate or have incomplete elementary education.