Klasifikasi Human Stress Menggunakan Adagrad Optimization untuk Arsitektur Deep Neural Network

Abstract

According to the World Health Organization, stress is a type of mental illness that affects human health and there is no one in this world who does not suffer from stress or depression. Stress is a term that is often used synonymously with negative life experiences or life events. . Analysis of data that has an unbalanced class results in inaccuracies in predicting human stress. This study shows that using the Deep Neural Network (DNN) Architecture model by optimizing several parameters, namely the optimizer, Learning rate and epoch. The best DNN Architect results are obtained with 4 Hidden Layers, Adagard Optimization, Learning rate 0.01 and the number of epochs 100. Accuracy, precision, recall and f-measure scores get 98.25%, 83.00%, 98.25%, 91.00%, respectively.