PTBA STOCK PRICE PREDICTION USING BIDIRECTIONAL LONG SHORT-TERM MEMORY WITH WHALE OPTIMIZATION ALGORITHM

Kumala, Pramudya Reksha (2026) PTBA STOCK PRICE PREDICTION USING BIDIRECTIONAL LONG SHORT-TERM MEMORY WITH WHALE OPTIMIZATION ALGORITHM. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 22081010186.-cover.pdf Download (2MB)
	Text (Bab 1) 22081010186.-bab1.pdf Download (176kB)
	Text (Bab 2) 22081010186.-bab2.pdf Restricted to Repository staff only until 22 July 2029. Download (480kB)
	Text (Bab 3) 22081010186.-bab3.pdf Restricted to Repository staff only until 22 July 2029. Download (674kB)
	Text (Bab 4) 22081010186.-bab4.pdf Restricted to Repository staff only until 22 July 2029. Download (949kB)
	Text (Bab 5) 22081010186.-bab5.pdf Download (167kB)
	Text (Daftar Pustaka) 22081010186.-daftarpustaka.pdf Download (133kB)
	Text (Lampiran) 22081010186.-lampiran.pdf Restricted to Repository staff only until 22 July 2029. Download (153kB)

Abstract

The stock price of PT Bukit Asam Tbk (PTBA) fluctuates due to global coal price movements and macroeconomic conditions, requiring an accurate prediction method to support investment decisions. Since PTBA's historical stock price data is a time series, a Bidirectional Long Short-Term Memory (BiLSTM) architecture was employed to capture its movement patterns. However, BiLSTM's performance is highly dependent on proper hyperparameter selection. Therefore, this study applies the Whale Optimization Algorithm (WOA) to search for the best hyperparameter combination, namely learning rate, hidden units, and dropout rate, using validation RMSE as the fitness function. The study uses 10 years of daily PTBA historical data (2016–2026), totaling 2,417 records obtained from Investing.com, using the Price, High, and Low features with an 80:10:10 data split ratio. The test results show that the BiLSTM model with a window size of 15, 2 hidden layers, and WOA with a population of 20 individuals and 40 iterations achieved an MAE of 29.59, RMSE of 49.72, and MAPE of 1.17%. These results outperform the LSTM model (MAPE 1.58%) and BiLSTM without optimization (MAPE 1.31%). These findings prove that WOA optimization significantly improves BiLSTM's prediction accuracy in forecasting PTBA stock prices.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Sari, Anggraini Puspita	NIDN0716088605	anggraini.puspita.if@upnjatim.ac.id
Thesis advisor	Nurlaili, Afina Lina	NIDN0013129303	afina.lina.if@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.87 Neural computers

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Pramudya Reksha Kumala

Date Deposited:

22 Jul 2026 06:40

Last Modified:

22 Jul 2026 07:42

URI:

https://repository.upnjatim.ac.id/id/eprint/57244