Prediksi Prestasi Akademik Mahasiswa Bekerja Paruh Waktu Menggunakan Artificial Neural Network

Abstract

Abstrack – Students who work part-time are required to be able to divide their time effectively and efficiently between time for work and time for study. The prediction of those who study while working is expected to be one of the policy considerations for the academic side so that students who work while working can complete their study period on time. This research begins with the stage of collecting data from students who are studying while working for the next data cleaning process. The data is then divided into two groups of data, namely training data and testing data which are normalized by the min-max method. Neural network algorithms are used to predict the results of studies for those who study while working which are categorized into 3 labels. Optimization is carried out on the parameters by utilizing the optimize parameter tool. In model testing, the parameters displayed are training cycle, learning rate, momentum, accuracy and RMSE value with a range of learning rate and momentum values from 0.1 to 0.9, with a sigmoid activation function. The best value validation was obtained in the training cycle 201, learning rate 0.74, momentum 0.9 with an accuracy value of 89.62%, RMSE 0.263 with a value of k-fold = 3.