MULTI-COMMODITY NON-OIL EXPORT FORECASTING IN INDONESIA USING BACKPROPAGATION ARTIFICIAL NEURAL NETWORK
DOI:
https://doi.org/10.54314/jssr.v8i4.4797Abstract
Abstract: Forecasting non-oil export commodities is critical for Indonesia's trade strategy, as these commodities contribute 93.8% of total national exports. This study develops a multi-commodity export prediction system using Artificial Neural Network (ANN) with backpropagation algorithm for 32 Indonesian non-oil commodities across six strategic sectors. Using monthly export data from February to August 2025 from Indonesia's Central Statistics Agency, we identified optimal neural network architecture 6-5-1 (6 input neurons for 6-month historical data, 5 hidden neurons, 1 output neuron). The model achieved 89.16% training accuracy and 88.43% testing accuracy with minimal 0.73% differential, indicating strong generalization without overfitting. Highest accuracy occurred on stable commodities (Tobacco: 99.94%, Animal/Plant Fats: 99.90%) while volatile commodities showed lower accuracy (Oil Seeds: 42.57%). The developed web-based system enables policymakers and exporters to make strategic decisions for international trade. This research demonstrates ANN backpropagation effectiveness for multi-dimensional commodity forecasting and provides practical decision-support tools for Indonesia's non-oil export sector.
Keyword: artificial neural network; backpropagation; export forecasting; commodity prediction; Indonesia.
Abstrak: Peramalan komoditas ekspor nonmigas sangat penting untuk strategi perdagangan Indonesia karena berkontribusi 93,8% dari total ekspor nasional. Penelitian ini mengembangkan sistem prediksi ekspor multi-komoditas menggunakan Jaringan Syaraf Tiruan (JST) dengan algoritma backpropagation untuk 32 komoditas nonmigas Indonesia di enam sektor strategis. Menggunakan data ekspor bulanan Februari-Agustus 2025 dari Badan Pusat Statistik Indonesia, kami mengidentifikasi arsitektur jaringan optimal 6-5-1 (6 neuron input untuk data 6 bulan, 5 neuron tersembunyi, 1 neuron output). Model mencapai akurasi training 89,16% dan testing 88,43% dengan diferensial minimal 0,73%, menunjukkan generalisasi kuat tanpa overfitting. Akurasi tertinggi pada komoditas stabil (Tembakau: 99,94%, Lemak dan Minyak: 99,90%) sedangkan komoditas volatil menunjukkan akurasi lebih rendah (Biji-bijian Berminyak: 42,57%). Sistem berbasis web memungkinkan pembuat kebijakan dan eksportir membuat keputusan strategis untuk perdagangan internasional. Penelitian ini menunjukkan efektivitas JST backpropagation untuk peramalan komoditas multi-dimensi dan menyediakan alat pengambilan keputusan praktis untuk sektor ekspor nonmigas Indonesia.
Kata kunci: jaringan syaraf tiruan; backpropagation; peramalan ekspor; prediksi komoditas; Indonesia.
Downloads
References
Badan Pusat Statistik. (2024). Statistik perdagangan luar negeri Indonesia 2023: Analisis komoditas ekspor non-migas. Jakarta: Badan Pusat Statistik Republik Indonesia.
Ben Ameur, H., Boubaker, S., Ftiti, Z., Louhichi, W., & Tissaoui, K. (2023). Forecasting commodity prices: Empirical evidence using deep learning tools. Annals of Operations Research, 339(1), 349-367.
Hidayat, A., & Rahman, F. (2022). Perancangan sistem prediksi berbasis neural network untuk optimasi strategi bisnis. Jurnal Teknologi dan Sistem Informasi, 15(4), 378-395.
Jin, X., & Xu, Y. (2024). Neural network models for international commodity price prediction with high volatility. International Journal of Forecasting, 40(2), 456-475.
Karasu, S., & Altan, A. (2022). Crude oil time series prediction model based on LSTM network with chaotic artificial bee colony algorithm. Energy, 239, 121-135.
Kementerian Perdagangan Republik Indonesia. (2024). Laporan analitik perdagangan internasional: Strategi diversifikasi ekspor non-migas. Jakarta: Kementerian Perdagangan.
Kusumadewi, S. (2018). Artificial intelligence: Teknik dan aplikasinya. Yogyakarta: Graha Ilmu.
Lewis, C. D. (1982). Industrial and business forecasting methods: A practical guide to exponential smoothing and curve fitting. London: Butterworths.
Murugesan, M., Krishnamurthy, R., & Devi, S. (2021). Neural networks and deep learning for agricultural commodity price prediction: Comparative study with ARIMA models. Computers and Agriculture, 8(4), 321-338.
Permana, I., & Salisah, F. N. (2022). The effect of data normalization on the performance of the classification. IJIRSE: Indonesian Journal of Informatic Research and Software Engineering, 2(1), 67-72.
Putri, A. D., Fitria, D., Amalita, N., & Zilrahmi. (2023). Implementation of backpropagation artificial neural network on forecasting export of palm oil in Indonesia. UNP Journal of Statistics and Data Science, 1(5), 473-479.
Siang, J. J. (2005). Jaringan syaraf tiruan & program menggunakan Matlab. Yogyakarta: Andi Publisher.
Siregar, M. R., Azhari, A. P., Hartama, D., & Windarto, A. P. (2022). Peramalan nilai penjualan gas elpiji 3 kg di Sumatera Utara dengan bantuan analisis metode jaringan syaraf
