Perbandingan Metode Hybrid CEEMDAN-SVR dan CEEMDAN-LSTM dalam Peramalan Harga Saham

Firdaus, Amira Zahra Anggraini (2026) Perbandingan Metode Hybrid CEEMDAN-SVR dan CEEMDAN-LSTM dalam Peramalan Harga Saham. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006221053-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Investasi saham dilakukan dengan tujuan untuk memperoleh keuntungan di masa depan sehingga memerlukan keputusan yang tepat agar risiko dapat diminimalkan. Peramalan harga saham menjadi langkah penting dalam pengambilan keputusan investasi. Akan tetapi, proses peramalan menghadapi tantangan akibat karakteristik data saham yang bersifat nonlinear dan mengandung noise yang tinggi. Penelitian ini bertujuan untuk membandingkan performa metode hybrid CEEMDAN-SVR dan CEEMDAN-LSTM dalam memprediksi harga saham PT Bank Rakyat Indonesia Tbk (BBRI) serta melakukan peramalan harga saham selama 20 hari menggunakan metode hybrid terbaik. Data yang digunakan berupa data historis harga penutupan saham harian BBRI periode 2 Januari 2023 hingga 30 Desember 2025. Proses dekomposisi menggunakan CEEMDAN menghasilkan 6 komponen IMF dan 1 komponen residual. Setiap komponen kemudian dimodelkan menggunakan Support Vector Regression (SVR) dan Long Short-Term Memory (LSTM) dengan beberapa skenario window size. Hasil prediksi masing-masing komponen kemudian direkonstruksi untuk memperoleh prediksi akhir lalu dievaluasi menggunakan Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), dan Mean Absolute Percentage Error (MAPE). Hasil penelitian menunjukkan bahwa metode hybrid CEEMDAN-LSTM dengan window size 9 memberikan performa terbaik dengan nilai MAE sebesar 21,815, RMSE sebesar 28,125, dan MAPE sebesar 0,567% mengungguli model terbaik dari CEEMDAN-SVR dengan window size 10 yang menghasilkan nilai MAE sebesar 25,085, RMSE sebesar 30,546, dan MAPE sebesar 0,649%. Selain itu, hasil peramalan harga saham BBRI menggunakan metode hybrid terbaik untuk 20 hari ke depan menunjukkan bahwa harga saham bergerak pada kisaran Rp 3.730 hingga Rp 3.870 tanpa menunjukkan perubahan tren yang ekstrem dalam jangka pendek.
========================================================================================================================
Stock investment aims to generate future returns. Therefore, appropriate investment decisions are required to minimize risks. Stock price forecasting plays an important role in supporting investment decision-making. However, the forecasting process remains a challenging task due to the nonlinear characteristics and high level of noise in stock price data. This study aims to compare the performance of the hybrid CEEMDAN-SVR and CEEMDANLSTM methods in predicting the stock price of PT Bank Rakyat Indonesia Tbk (BBRI) and to forecast its stock price for the next 20 trading days using the best-performing hybrid method. The data used in this study consist of the daily historical closing price of BBRI from January 2, 2023 to December 30, 2025. The decomposition process using CEEMDAN produces 6 IMF components and 1 residual component. Each component is then modeled separately using Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) with several window size scenarios. The predicted values of all components are reconstructed to obtain the final prediction and evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), dan Mean Absolute Percentage Error (MAPE). The results show that the hybrid CEEMDAN-LSTM model with window size 9 achieved the best performance, with MAE of 21,815, RMSE of 28,125, and MAPE of 0,567%, outperforming the best CEEMDAN-SVR model with window size 10 which obtained MAE of 25,085, RMSE of 30,546, and MAPE of 0,649%. Furthermore, the forecasting results for the next 20 days indicate that BBRI stock prices are projected to move within a range of approximately Rp3.730 to Rp3.870 without showing any extrem short-term trend changes.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	CEEMDAN, Harga Saham, LSTM, Peramalan, SVR, CEEMDAN, Forecasting, LSTM, Stock Price, SVR
Subjects:	H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. H Social Sciences > HG Finance > HG4915 Stocks--Prices Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Amira Zahra Anggraini Firdaus
Date Deposited:	16 Jul 2026 09:41
Last Modified:	16 Jul 2026 09:47
URI:	http://repository.its.ac.id/id/eprint/135124

Actions (login required)

View Item