Comparison of Classification Performances of Mathematics Achievement at PISA 2012 with the Artificial Neural Network, Decision Trees and Discriminant Analysis
| Autor: | Selahattin Gelbal, Emre Toprak |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Artificial Neural Netwok
Decision Trees Discriminant Analysis PISA Student Achievement Artificial neural network business.industry Computer Science::Neural and Evolutionary Computation 05 social sciences Decision tree 050401 social sciences methods 050301 education Eğitim Bilimsel Disiplinler Ocean Engineering Academic achievement Machine learning computer.software_genre Linear discriminant analysis Correlation ComputingMethodologies_PATTERNRECOGNITION 0504 sociology Sample size determination Achievement test Artificial intelligence business Education Scientific Disciplines 0503 education computer Mathematics |
| Zdroj: | Volume: 7, Issue: 4 773-799 International Journal of Assessment Tools in Education |
| ISSN: | 2148-7456 |
| Popis: | This study aims to compare the performances of the artificial neural network, decision trees and discriminant analysis methods to classify student achievement. The study uses multilayer perceptron model to form the artificial neural network model, chi-square automatic interaction detection (CHAID) algorithm to apply the decision trees method and linear discriminant analysis. The performance of each method has been investigated in different sample sizes when classifying into different numbered subgroups. The study has revealed that the artificial neural network has the best performance in large, medium and small sample sizes when classifying into six, three and two subgroups. In the very small sample size, which has homogeneous variance-covariance matrices, the discriminant analysis performs the best, while in the very small sample size, which does not have homogeneous variance-covariance matrices, it is the discriminant analysis which performs the best when classifying into six subgroups and the artificial neural network performs the best when classifying into two and three subgroups. Considering the performances of the methods with respect to sample size, it can be concluded that as the sample size gets smaller, the performance of the decision trees method gets worse, whereas the performance of the discriminant analysis method improves. No correlation of this kind has been found with regard to the artificial network method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|