KLIK (Kumpulan jurnaL Ilmu Komputer) (e-Journal)
Vol 10, No 1 (2023)

IMPLEMENTATION OF ARTIFICIAL NEURAL NETWORK (ANN) USING BACKPROPAGATION ALGORITHM BY COMPARING FOUR ACTIVATION FUNCTIONS IN PREDICTING GOLD PRICES

Dian Kurniasari (Universitas Lampung)
Ranti Vidia Mahyunis (Universitas Lampung)
Warsono Warsono (Universitas Lampung)
Aang Nuryaman (Universitas Lampung)



Article Info

Publish Date
28 Feb 2023

Abstract

The trend in global currency values is speedy and fluctuating due to the recession caused by the Covid-19 pandemic. That causes investors to flock to buy gold assets. Therefore, it is necessary to predict the price of gold from a business and academic perspective to obtain a reasonable gold price prediction model. This study applies the Backpropagation Algorithm by determining the best ANN model structure based on four activation functions: Sigmoid, Tanh, ReLU, and Linear, as well as learning rate values, namely 0.01 and 0.001. The results are the best ANN model structure with four nodes in the input layer, four nodes in the hidden layer and the output layer using the Linear activation function and a learning rate of 0.01. Based on the structure of the model, the MSE value is 0.00051, the MAPE value is 1.9798%, and the accuracy is 98%.Keywords: Artificial Neural Network, Backpropagation, Gold Price Prediction, Activation Function, Model Structure Trend nilai mata uang global sangat cepat dan fluktuatif akibat terjadinya resesi yang disebabkan oleh pandemi Covid-19. Hal ini menyebabkan, para investor berbondong-bondong untuk membeli aset emas. Oleh sebab itu, perlu dilakukan prediksi harga emas, baik dari perspektif bisnis maupun akademis agar memperoleh model prediksi harga emas yang baik. Penelitian ini menerapkan Algoritma Backpropagation dengan menentukan struktur model ANN terbaik berdasarkan empat fungsi aktivasi yaitu, Sigmoid, Tanh, ReLU, dan Linear serta nilai learning rate, yaitu 0,01 dan 0,001. Hasil yang diperoleh berupa struktur model ANN terbaik dengan empat node pada input layer, empat node pada hidden layer dan output layer dengan menggunakan fungsi aktivasi Linear dan learning rate sebesar 0,01. Berdasarkan struktur model tersebut, diperoleh nilai MSE sebesar 0.00051, nilai MAPE sebesar 1,9798% dan akurasi sebesar 98%.Kata Kunci: Artificial Neural Network, Backpropagation, Prediksi Harga Emas, Fungsi Aktivasi, Struktur Model

Copyrights © 2023






Journal Info

Abbrev

klik

Publisher

Subject

Computer Science & IT

Description

KLIK Scientific Journal, is a computer science journal as source of information in the form of research, the study of literature, ideas, theories and applications in the field of critical analysis study Computer Science, Data Science, Artificial Intelligence, and Computer Network, published two ...