Rangkaian Neural Untuk Peramalan Gred Matematik Tambahan

The student academic achievement in the Sijil Pelajaran Malaysia (SPM) has been the important measurement towards the quality of education in Malaysia’s school level. However, it is difficult to find an efficient measurement tool in order to assist teachers in evaluating their students’ performance before the SPM examination. The study attempts to develop a backpropagation neural networks
model to predict the students’ achievement grades in SPM examination for Additional Mathematics subject. The dataset were the year 1998’s SPM candidates collected from nineteen schools throughout Alor Setar, Pulau Pinang,
Ipoh and Shah Alam. These dataset are used to train and test the neural networks model, The model uses neural networks topology that consists of seven nodes of
the input layer, six nodes in the hidden layer and a node in the output layer. The obtained model is able to perform with 83.68% of correct prediction towards the test dataset, compared to 76.47% using the regression technique. Further work can be done on improving the data preprocessing technique and the network structure. The model can also be integrated with the database and knowledge based systems.