Ristiyani, Sintiya (2025) PERBANDINGAN MODEL MIXTURE AUTOREGRESSIVE (MAR) NORMAL DAN MAR-GED DALAM PREDIKSI HARGA SAHAM EMITEN ROKOK DI BURSA EFEK INDONESIA. Undergraduate thesis, Universitas Pembangunan Nasional "Veteran" Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
21083010103_Cover.pdf Download (1MB) |
|
Text (Bab 1)
21083010103_BAB 1.pdf Download (240kB) |
|
Text (Bab 2)
21083010103_BAB 2.pdf Restricted to Repository staff only until 25 July 2028. Download (488kB) |
|
Text (Bab 3)
21083010103_BAB 3.pdf Restricted to Repository staff only until 25 July 2028. Download (490kB) |
|
Text (Bab 4)
21083010103_BAB 4.pdf Restricted to Repository staff only until 25 July 2028. Download (1MB) |
|
Text (Bab 5)
21083010103_BAB 5.pdf Restricted to Repository staff only Download (224kB) |
|
Text (Daftar Pustaka)
21083010103_Daftar Pustaka.pdf Download (199kB) |
|
Text (Lampiran)
21083010103_Lampiran.pdf Restricted to Repository staff only until 25 July 2028. Download (501kB) |
Abstract
Stock price prediction plays a crucial role in investment decision-making in capital markets. One statistical approach capable of capturing complex characteristics of financial data such as volatility and non-normal Return distributions is the Mixture Autoregressive (MAR) model. This study aims to compare the performance of two MAR model variants, namely MAR-Normal and MAR-GED (Generalized Error Distribution), in forecasting the stock prices of tobacco companies listed on the Indonesia Stock Exchange, including PT Gudang Garam Tbk (GGRM), PT HM Sampoerna Tbk (HMSP), and PT Wismilak Inti Makmur Tbk(WIIM). The dataset consists of daily closing prices from August 2021 to August 2024, which are transformed into log Returns and tested for stationarity. Model development involves selecting the optimal model structure based on the Bayesian Information Criterion (BIC) and estimating parameters using the Expectation-Maximization (EM) algorithm. Model performance is evaluated through AR parameter significance testing, residual diagnostic analysis, and 30 step ahead forecast accuracy using the Mean Absolute Percentage Error (MAPE) metric. The results indicate that, in general, the MAR-GED model outperforms the MAR-Normal model, particularly in capturing the heavy-tailed and asymmetric nature of stock Return distributions. MAR-GED also yields lower MAPE values in most of the stocks analyzed. Therefore, MAR-GED is deemed more effective for modeling and forecasting the stock price movements of tobacco issuers in Indonesia.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | H Social Sciences > HA Statistics Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Sintiya Ristiyani | ||||||||||||
| Date Deposited: | 28 Jul 2025 07:27 | ||||||||||||
| Last Modified: | 28 Jul 2025 07:27 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/40867 |
Actions (login required)
 |
View Item |