Analisa Jaringan Saraf Tiruan Backpropagation Untuk Memprediksi Prestasi Siswa SMA Muhammadiyah Serbelawan

Abstract

Achievements achieved by graduates from an educational institution show the quality and quality. One of them is seen from one of the assessment criteria for assessing the achievement of graduates at the secondary school level, namely through the average score. This average value is often used as a measure to assess students who will enter the next level of education. In addition, the acceptance of students at a level of education is also adjusted to the capacity of the school in question. The high average score at the high school level does not guarantee student achievement at the tertiary level. So that this study aims to obtain an output architecture prediction of student achievement at SMA Muhammadiyah Serbelawan which correlates between the average value and the total score of class XII (twelve) high school students according to the data trained using Artificial Neural Network Analysis using the Backpropagation method. The data taken in the form of the average value of students and the total value of the second semester of class XII students. Furthermore, the data were analyzed using Backpropagation ANN method, with the help of MATLAB software. From the results of testing the Student Achievement data above, we can see in the 5-5-5-1 architecture which shows from the target minus the ANN output that SSE is 0.17625 which shows that there is a measuring tool in predicting the best students using academic value data as a target. From the data obtained, the computational performance of artificial neural networks with the Backpropagation Algorithm is 85%.