Peramalan Harga Crude Coconut Oil Menggunakan Multihead Attention-Long Short Term Memory (Multihead Att-Lstm)

Siahaan, Thasya Rosa Chrysisosiarini (2024) Peramalan Harga Crude Coconut Oil Menggunakan Multihead Attention-Long Short Term Memory (Multihead Att-Lstm). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06111840000093-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 July 2026. Download (1MB) | Request a copy

Abstract

Crude Coconut Oil (CCO) menjadi salah satu olahan perkebunan kelapa berupa minyak mentah kelapa yang menjadi salah satu komoditas ekspor di Indonesia. Pada tahun 2017, Indonesia menempati peringkat pertama negara penghasil kelapa, dengan produksi mencapai 18,9 juta ton. CCO merupakan salah satu bahan baku penting dalam industri pengolahan makanan.Di pasar dunia, harga CCO cenderung fluktuatif, sehingga diperlukan peramalan untuk mengetahui harga CCO pada waktu yang akan datang. Pada penelitian ini telah dilakukan penerapan metode MultiHead Attention dan Long Short-Term Memory dimana MultiHead Attention merupakan sebuah mekanisme yang digunakan pada struktur transformer dan LSTM merupakan sebuah jenis arsitektur Recurrent Neural Network (RNN) yang dirancang untuk menangani permasalahan yang terkait dengan pembelajaran sequence jangka panjang. Penelitian ini menggunakan data harian harga CCO mulai dari tahun 2007 hingga 2024. Hasil penelitian menunjukkan bahwa algoritma MultiHead Attention-LSTM mampu meramalkan harga CCO dengan baik. Hal ini dibuktikan dengan diperolehnya nilai Mean Absolute Error (MAE) sebesar 36.9894, nilai Mean Absolute Percentage Error (MAPE) sebesar 2.3677%, nilai Mean Square Error (MSE) sebesar 4170.7324, dan nilai Root Mean Square Error (RMSE) sebesar 64.5812. Hasil ujicoba juga menunjukkan bahwa algoritma MultiHead Attention-LSTM lebih baik daripada algoritma LSTM
===================================================================================================================================
Crude Coconut Oil (CCO) is one of the processed coconut plantations in the form of crude coconut oil which is one of the export commodities in Indonesia. In 2017, Indonesia was ranked first as a coconut producing country, with production reaching 18.9 million tons. CCO is an important raw material in the food processing industry. On the world market, CCO prices tend to fluctuate, so forecasting is needed to know the CCO price in the future. In this research, the MultiHead Attention and Long Short-Term Memory methods have been applied, where MultiHead Attention is a mechanism used in transformer structures and LSTM is a type of Recurrent Neural Network (RNN) architecture designed to handle problems related to long-term sequence learning. . This research uses daily CCO price data from 2007 to 2024. The research results show that the MultiHead Attention-LSTM algorithm is able to predict CCO prices well. This is proven by obtaining a Mean Absolute Error (MAE) value of 36.9894, a Mean Absolute Percentage Error (MAPE) value of 2.3677%, a Mean Square Error (MSE) value of 4170.7324, and a Root Mean Square Error (RMSE) value of 64.5812. The test results also show that the MultiHead Attention-LSTM algorithm is better than the LSTM algorithm

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Crude Coconut Oil (CCO), Forecasting, MultiHead Attention, Long Short-Term Memory (LSTM), Time Series
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Thasya Rosa Chrysisosiarini Siahaan
Date Deposited:	19 Feb 2024 07:06
Last Modified:	19 Feb 2024 07:06
URI:	http://repository.its.ac.id/id/eprint/107487

Actions (login required)

View Item