Loading...
Publication
Algoritmo cart : previsão do desempenho na matemática do secundário
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 153.01 KB | Adobe PDF |
Advisor(s)
Abstract(s)
O algoritmo CART-Classification and Regression Trees é aplicado na previsão das
classificações de matemática associadas a uma amostra de alunos do ensino secundário. São
modeladas, separadamente, as observações respeitantes a alunos do ensino secundário público
e privado, considerando factores sócio-demográficos, factores específicos e factores pessoais.
Obtém-se uma boa capacidade preditiva para os modelos propostos: 83,5% e 90,5%,
estimativas da proporção de variância explicada, obtidas mediante validação cruzada, para os
modelos do ensino público e privado, respectivamente. É ainda avaliada a importância
relativa das variáveis preditivas nos modelos sublinhando-se, como principal, a média obtida pelos alunos às restantes disciplinas do secundário.
Abstract: In the present study we use the CART-Classification and Regression Trees algorithm to predict math grades based on a sample of high school students. Students from Public and Private schools are considered separately. Predictors include socio-demographics, personal attributes and some specific characteristics related to school. The models obtained have a good predictive capacity: proportion of grades’ explained variance is 83,5% and 90,5% for regression trees referred to Public and Private schools, respectively. The relative importance of predictors is evaluated, the most important being the student’s average grade referred to the remaining subjects (excluding mathematics).
Abstract: In the present study we use the CART-Classification and Regression Trees algorithm to predict math grades based on a sample of high school students. Students from Public and Private schools are considered separately. Predictors include socio-demographics, personal attributes and some specific characteristics related to school. The models obtained have a good predictive capacity: proportion of grades’ explained variance is 83,5% and 90,5% for regression trees referred to Public and Private schools, respectively. The relative importance of predictors is evaluated, the most important being the student’s average grade referred to the remaining subjects (excluding mathematics).
Description
Keywords
Árvores de regressão Algoritmo CART Previsão Ensino das matemáticas Regression trees CART algorithm Prediction Teaching of mathematics
Citation
Cabete, Nélia Pereira; Cardoso, Margarida G. M. S. - Algoritmo cart : previsão do desempenho na matemática do secundário. "Revista de Ciências da Computação" [Em linha]. ISSN 1646-6360. Vol. 1, nº1 (2006), p. 27-55
Publisher
Universidade Aberta