Peramalan Harga Nikel Indonesia Menggunakan Metode Long Short-Term Memory

Oktavia, Afina Zhahira (2024) Peramalan Harga Nikel Indonesia Menggunakan Metode Long Short-Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003201057-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Adanya kemajuan teknologi yang mengarah pada transisi energi ramah lingkungan, menyebabkan produsen kendaraan bermotor mulai berkompetisi membuat kendaraan bertenaga listrik. Salah satu bahan yang umum digunakan dalam proses pembuatan baterai kendaraan listrik adalah nikel. Peramalan harga nikel diperlukan untuk penetapan kebijakan perusahaan seperti penentuan harga jual dan jumlah pembelian bahan baku. Peramalan harga nikel di Indonesia sendiri menjadi salah satu tantangan mengingat harga nikel memiliki karakteristik yang sangat tidak stabil dan kompleks. Karakteristik harga nikel tersebut membuat pihak yang berkaitan seperti perusahaan, investor, dan kensumen mengalami kesulitan dalam memanfaatkan peluang bisnis, operasi, dan konsumsi secara akurat. Dalam penelitian ini digunakan metode LSTM yang memiliki kemampuan baik dalam meramalkan data nonlinear jika dibandingkan dengan metode peramalan konvensional seperti ARIMA dan neural network. Tujuan utama dari penelitian ini adalah untuk mengetahui performa dari model LSTM dalam meramalkan harga nikel Indonesia serta mendapatkan hasil peramalannya. Dari proses analisis dan pembahasan diperoleh model LSTM terbaik dengan jumlah hidden layer 1 dan neuron 60. Model tersebut menghasilkan nilai rata-rata AIC in-sample 26722,61378 dan rata-rata RMSE out-sample 9761,42336. Pada peramalan data out-sample secara k-step model LSTM menghasilkan nilai sMAPE sebesar 1,385%, sedangkan pada peramalan data out-sample secara 1-step model LSTM menghasilkan nilai sMAPE sebesar 0,404%. Hasil peramalan harga nikel Indonesia menggunakan LSTM menunjukkan pola stasioner atau mendatar yakni pada rentang harga Rp 256.432,47 sampai dengan Rp 266.975,20 untuk 20 periode kedepan.
==================================================================
Technological advances that lead to an environmentally friendly energy transition have caused motor vehicle manufacturers to start competing in making electric-powered vehicles. One of the materials commonly used in the process of making electric vehicle batteries is nickel. Nickel price forecasting is needed to determine company policies such as determining selling prices and the amount of raw materials purchased. Forecasting nickel prices in Indonesia itself is a challenge considering that nickel prices have very unstable and complex characteristics. The characteristics of nickel prices make it difficult for related parties such as companies, investors and consumers to take advantage of business, operational and consumption opportunities accurately. In this research, the LSTM method is used which has good capabilities in predicting nonlinear data when compared to conventional forecasting methods such as ARIMA and neural networks. The main objective of this research is to determine the performance of the LSTM model in predicting Indonesian nickel prices and to obtain forecasting results. From the analysis and discussion process, the best LSTM model was obtained with the number of hidden layers 1 and 60 neurons. This model produced an average in-sample AIC value of 26722.61378 and an average out-sample RMSE of 9761.42336. When forecasting out-sample data using k-step, the LSTM model produces an sMAPE value of 1.385%, while when forecasting out-sample data using 1-step, the LSTM model produces an sMAPE value of 0.404%. The results of forecasting Indonesian nickel prices using LSTM show a stationary or horizontal pattern, namely in the price range of IDR 256,432.47 to IDR 266,975.20 for the next 20 periods.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	LSTM, Nickel, Nonlinear, LSTM, Nikel, Nonlinear
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Afina Zhahira Oktavia
Date Deposited:	08 Aug 2024 09:03
Last Modified:	08 Aug 2024 09:03
URI:	http://repository.its.ac.id/id/eprint/114915

Actions (login required)

View Item