Comparing the Predictive and Classification Performances of Logistic Regression and Neural Networks: A Case Study on Timss 2011☆

摘要: Abstract Investigating effective factors on students’ achievement has wide application area in educational studies. Specially, Trends International Mathematics and Science Study (TIMSS) allows researchers to determine correlates of mathematics science for different countries. In this study, the predictive classification performances logistic regression neural networks are compared identify impact levels variables Turkey. Age, gender scales created by TIMSS team 8th grade students (students like learning, value confident math, engaged bullied at school, home resources), selected as variables. Model fitting statistics show that two methods give similar results prediction classification. addition model results, confidence is found most factor improve achievement.