Peramalan Harga Saham Menggunakan Metode Hybrid LSTM-GRU (Studi Kasus: PT Indofood CBP Sukses Makmur Tbk.)

Setiawan, Fikri Septa (2025) Peramalan Harga Saham Menggunakan Metode Hybrid LSTM-GRU (Studi Kasus: PT Indofood CBP Sukses Makmur Tbk.). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5026211109-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (10MB) | Request a copy

Abstract

Investasi saham merupakan salah satu instrumen keuangan yang diminati karena potensi keuntungan melalui dividen dan capital gain. Salah satu saham unggulan di Indonesia adalah PT Indofood CBP Sukses Makmur Tbk (ICBP), yang memiliki kapitalisasi besar dan likuiditas tinggi. Namun, harga saham ICBP dipengaruhi oleh volatilitas pasar, sehingga diperlukan strategi peramalan yang akurat. Penelitian ini mengembangkan model peramalan harga saham ICBP menggunakan metode hybrid Long Short-Term Memory (LSTM) dan Gated Recurrent Unit (GRU), yang menggabungkan keunggulan kedua metode. LSTM mampu menangani ketergantungan jangka panjang dalam data time-series, sementara GRU lebih efisien dalam komputasi. Berdasarkan eksperimen yang dilakukan, konfigurasi terbaik yang diperoleh dalam penelitian ini adalah sequence length sebesar 9, 256 unit LSTM, 128 unit GRU, batch size 32, learning rate 0,001, optimizer Adam, serta pembagian data 80% untuk pelatihan dan 20% untuk pengujian. Model hybrid LSTM-GRU yang dikembangkan mampu memberikan hasil prediksi yang akurat dengan nilai Mean Absolute Percentage Error (MAPE) sebesar 1,205% pada data dengan interpolasi outlier. Hasil ini menunjukkan bahwa model dapat memahami pola pergerakan saham dengan baik dan bahwa penggunaan interpolasi outlier dapat meningkatkan akurasi prediksi. Diharapkan model ini dapat membantu investor dalam mengambil keputusan investasi yang lebih tepat dengan memahami tren harga saham secara lebih presisi.
==========================================================================================================================================
Stock investment is one of the most favoured financial instruments due to its profit potential through dividends and capital gains. One of the leading stocks in Indonesia is PT Indofood CBP Sukses Makmur Tbk (ICBP), which has a large capitalisation and high liquidity. However, ICBP's share price is affected by market volatility, so an accurate forecasting strategy is needed. This research develops a stock price forecasting model for ICBP using a hybrid Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) method, which combines the advantages of both methods. LSTM is able to handle long-term dependencies in time-series data, while GRU is more computationally efficient. Based on the experiments conducted, the best configuration obtained in this study is a sequence length of 9, 256 LSTM units, 128 GRU units, batch size 32, learning rate 0.001, optimizer Adam, and data split 80% for training and 20% for testing. The hybrid LSTM-GRU model developed is able to provide accurate prediction results with a Mean Absolute Percentage Error (MAPE) value of 1.205% on data with outlier interpolation. These results show that the model can understand the stock movement pattern well and that the use of outlier interpolation can improve the prediction accuracy. It is hoped that this model can help investors make more informed investment decisions by understanding stock price trends with greater precision.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Investasi, Saham, Long Short-Term Memory, Gated Recurrent Unit, Peramalan, Investment, Stocks, Long Short-Term Memory, Gated Recurrent Unit, Forecasting
Subjects:	H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Fikri Septa Setiawan
Date Deposited:	24 Jul 2025 03:00
Last Modified:	24 Jul 2025 03:00
URI:	http://repository.its.ac.id/id/eprint/121113

Actions (login required)

View Item