Machine Learning untuk Prediksi Produksi Gula Nasional
Abstract
Gula merupakan salah satu bahan utama yang selalu dibutuhkan untuk membuat makanan atau minuman. Saat ini produksi gula belum mampu memenuhi kebutuhan gula nasional. Namun, pemerintah terus menambah jumlah pabrik dan meningkatkan produktivitas pabrik gula yang ada untuk memenuhi kebutuhan gula nasional. Di sisi lain, produksi gula juga berfluktuasi. Hal ini dapat mengakibatkan kelangkaan gula jika tidak diprediksi secara tepat dan akurat. Oleh karena itu, pada penelitian ini dilakukan perbandingan prediksi produksi gula Indonesia dengan menggunakan metode machine learning yaitu Long Short Term Memory (LSTM) dan juga dilakukan prediksi dengan menggunakan metode regresi linier. Penelitian ini dilakukan berdasarkan data sekunder yang bersumber dari hasil penelitian dan laporan dari instansi atau instansi terkait. Data yang digunakan adalah data nasional berupa time series selama 52 tahun yaitu dari tahun 1968 sampai dengan tahun 2020. Hasil penelitian ini menunjukkan bahwa error pada metode regresi linier adalah 8%. Hasil prediksi dengan machine learning menunjukkan error yang lebih kecil dibandingkan dengan metode regresi linier. Metode LSTM menghasilkan error data kereta sebesar 0,069% dan nilai error data pengujian sebesar 0,082%. Hasil peramalan dari regresi linier memiliki trend produksi yang meningkat namun pada metode LSTM hasilnya mengalami trend penurunan.
Sugar is one of the main ingredients that are always needed to make food or drink. At this time sugar production has not been able to meet the national sugar demand. However, the government continues to increase the number of factories and increase the productivity of existing sugar factories to meet national sugar demand. On the other hand, sugar production also fluctuated. This can lead to a shortage of sugar if it is not predicted precisely and accurately. Therefore, in this study, a comparison of predictions of Indonesian sugar production was carried out using the machine learning method, namely Long Short Term Memory (LSTM) and predictions were also made using a linear regression method. This research was conducted based on secondary data sourced from research results and reports from related institutions or agencies. The data used is national data in the form of a time series for 52 years, namely from 1968 to 2020. The results of this study show that the error in the linear regression method is 8%. Prediction results with machine learning show a smaller error than the linear regression method. The LSTM method produces a train data error of 0.069% and a test data error value of 0.082%. Forecasting results from linear regression have an increasing trend in production but in the LSTM method the results experience a downward trend.
Keywords
Full Text:
PDFReferences
B. P. Statistik, Distribusi Perdagangan Komoditas Gula Pasir. 2021.
M. Z. Hassan, M. Siraj-Ud-Doulah, and M. K. Hasan, “Forecasting The Production of Sugar Cane Based on Time series Models in Bangladesh,” Bull. Math. Stat. Res., vol. 7, no. 4, pp. 77–89, 2019, doi: 10.33329/bomsr.74.24.
A. K. Srivastava and M. K. Rai, “Sugarcane production: Impact of climate change and its mitigation,” Biodiversitas, J. Biol. Divers., vol. 13, no. 4, pp. 214–227, 2012, doi: 10.13057/biodiv/d130408.
D. Zhao and Y. R. Li, “Climate Change and Sugarcane Production: Potential Impact and Mitigation Strategies,” Int. J. Agron., vol. 2015, 2015, doi: 10.1155/2015/547386.
A. A. Sulaiman, Y. Sulaeman, N. Mustikasari, D. Nursyamsi, and A. M. Syakir, “Increasing Sugar Production in Indonesia through Land Suitability Analysis and Sugar Mill Restructuring,” MDPI, vol. 8, no. 61, pp. 1–17, 2019.
A. Kantasa-ard, M. Nouiri, A. Bekrar, A. Ait el cadi, and Y. Sallez, “Machine learning for demand forecasting in the physical internet: a case study of agricultural products in Thailand,” Int. J. Prod. Res., vol. 59, no. 24, pp. 7491–7515, 2021, doi: 10.1080/00207543.2020.1844332.
I. Harlianingtyas, A. Salim, D. Hartatie, and S. Supriyadi, “Forecasting sugarcane production in the Asembagus sugar factory,” in IOP Conference Series: Earth and Environmental Science, 2020, vol. 411, no. 1, p. 12022, doi: 10.1088/1755-1315/411/1/012022.
F. N. Fauziah and A. Gunaryati, “Comparison forecasting with double exponential smoothing and artificial neural network to predict the price of sugar,” Int. J. Simul. Syst. Sci. Technol., vol. 18, no. 4, pp. 11–13, 2017, doi: 10.5013/IJSSST.a.18.04.13.
K. Chen, Y. Zhou, and F. Dai, “A LSTM-based method for stock returns prediction: A case study of China stock market,” Proc. - 2015 IEEE Int. Conf. Big Data, IEEE Big Data 2015, pp. 2823–2824, 2015, doi: 10.1109/BigData.2015.7364089.
M. A. Heryanto and E. R. Suryatmana, “Dinamika Agroindustri Gula Indonesia: Tinjauan Analisis Sistem,” Agricore J. Agribisnis dan Sos. Ekon. Pertan. Unpad, vol. 5, no. 2, 2020.
J. Christy, A. P. Hintarsyah, and H. L. H. Spits Warnars, “Forecasting Sebagai Decision Support Systems Aplikasi dan Penerapannya Untuk Mendukung Proses Pengambilan Keputusan.,” J. Sist. Komput., vol. 8, no. 1, pp. 19–27, 2018.
DOI: https://doi.org/10.18196/jmpm.v6i1.14897
Refbacks
- There are currently no refbacks.
Editorial Office :
JMPM (Jurnal Material dan Proses Manufaktur)
Department of Mechanical Engineering, Faculty of Engineering, Universitas Muhammadiyah Yogyakarta.
Jl. Brawijaya Tamantirto Kasihan Bantul 55183 Indonesia
Email: jmpm@umy.ac.id
Telp/Fax: (62)274-387656
Whatsapp: 085642095827