Proceedings of The 4th International Conference on Technology Education in the Asia-Pacific Region (ICTE),pp.33-43(2001), Deajon, Jorea

Development of an Evaluation System Using Neural Networks
for the Learning of Computer Programming



@ In this paper, the possibility of the application of Neural Networks to evaluation of classroom instruction in computer programming (Logo) is discussed.
An Evaluation system using Back Propagation Neural Networks with Bias Units was designed by learning simulation. The accuracy of this system was evaluated by recognition simulation. The results showed, the correlation between measured and predicted scores obtained in learning simulation was 0.821, and the absolute value of error was under 0.1 in 89.1% of the total of predicted scores in recognition simulation. Based on these scores, an evaluation score set was constructed. This evaluation score set was used for analysis of the effects of practical instruction.
As a result, the development processes of students' programming abilities could be evaluated, and it was indicated that case studies of debugging and the collaborative programming had a marked effect on their programming abilities.

Keywords: Evaluation System, Neural Network, Logo Programming, and Junior high school