Lumangkun, Mordekhai Gerine (2026) OPTIMIZING HYBRID LSTM-GRU HYPERPARAMETERS USING GENETIC ALGORITHM FOR PREDICTING BANKING STOCK PRICES LISTED IN THE LQ45 INDEX. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (Cover)
20081010041-cover.pdf Download (6MB) | Preview |
|
|
Text (BAB I)
20081010041-bab1.pdf Download (115kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (BAB II)
20081010041-bab2.pdf Restricted to Repository staff only until 4 May 2028. Download (448kB) |
|
|
Text (BAB III)
20081010041-bab3.pdf Restricted to Repository staff only until 4 May 2028. Download (1MB) |
|
|
Text (BAB IV)
20081010041-bab4.pdf Restricted to Repository staff only until 4 May 2028. Download (3MB) |
|
|
Text (BAB V)
20081010041-bab5.pdf Download (132kB) | Preview |
|
|
Text (Daftar Pustaka)
20081010041-daftarpustaka.pdf Download (99kB) | Preview |
|
|
Text (Lampiran)
20081010041-lampiran.pdf Restricted to Repository staff only until 4 May 2028. Download (771kB) |
Abstract
This study is motivated by the high volatility of banking sector stock prices in the LQ45 index, which calls for accurate and adaptive forecasting methods. Advances in deep learning, particularly Recurrent Neural Network (RNN) models such as Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), provide the capability to capture patterns in time-series data. However, model performance is highly influenced by the selection of optimal hyperparameters. Therefore, this study examines the application of a hybrid LSTM-GRU model optimized using a Genetic Algorithm (GA). The data used includes daily stock prices and internal company factors such as ROA and ROE for the period from November 2020 to June 2025. The research stages include collecting daily stock price data and financial ratio reports from 6 banking issuers listed in the LQ45 Index, data preprocessing, model training, and hyperparameter optimization using GA. Model evaluation was conducted using RMSE, MAE, and MAPE. The results show that the LSTM-GRU hybrid model optimized with GA significantly improves prediction accuracy compared to the unoptimized model, as evidenced by a cumulative reduction in the average error values across all evaluation metrics: 18.42% for RMSE, 20.61% for MAE, and 19.80% for MAPE. Thus, the combination of the hybrid method and evolutionary-based optimization has proven effective in improving stock price prediction performance.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Unnamed user with email 20081010041@student.upnjatim.ac.id | ||||||||||||
| Date Deposited: | 04 May 2026 08:15 | ||||||||||||
| Last Modified: | 04 May 2026 08:15 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/51486 |
Actions (login required)
 |
View Item |