Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/166412| Title: | Penggerombolan Data Deret Waktu untuk Peramalan Harga Saham Sektor Consumer Non Cyclicals |
| Other Titles: | |
| Authors: | Anisa, Rahma Wijayanto, Hari Rahmi, Salsabila Dwi |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Meningkatnya jumlah investor di Indonesia mencerminkan minat yang terus bertumbuh terhadap instrumen pasar modal, khususnya pada sektor consumer non cyclicals yang dikenal stabil dan tahan terhadap fluktuasi ekonomi. Penelitian ini bertujuan untuk: (1) membandingkan kinerja penggerombolan data harga saham sektor consumer non cyclicals menggunakan ukuran ketidakmiripan Complexity Invariant Distance (CID) dan Dynamic Time Warping (DTW), (2) mengetahui karakteristik pergerakan harga saham pada setiap gerombol yang terbentuk, dan (3) melakukan peramalan harga saham pada level gerombol menggunakan metode Triple Exponential Smoothing (Holt-Winter’s) multiplikatif dan aditif. Pemilihan DTW didasarkan pada kemampuannya menyelaraskan deret waktu dengan panjang atau pola yang berbeda, sementara CID dipilih karena sensitif terhadap kompleksitas bentuk data deret waktu. Data yang digunakan berupa harga penutupan mingguan dari 129 emiten sektor consumer non cyclicals yang terdaftar di Bursa Efek Indonesia selama periode 1 Januari 2019 hingga 30 Desember 2024. Hasil penelitian menunjukkan bahwa DTW dengan pautan rataan pada metode penggerombolan aglomeratif berhirarki menghasilkan penggerombolan terbaik, dengan empat gerombol optimum dan koefisien silhouette sebesar 0,75, yang menunjukkan kualitas pengelompokan yang kuat. Peramalan menggunakan HoltWinter’s menunjukkan kinerja prediksi sangat baik dengan nilai MAPE di bawah 10%, yakni 4,62% untuk Gerombol A, 3,24% untuk Gerombol B, 3,06% untuk Gerombol C, dan 3,82% untuk Gerombol D. Temuan ini memberikan pendekatan yang efisien dan akurat dalam memahami pola pergerakan harga saham serta mendukung pengambilan keputusan investasi pada sektor consumer non cyclicals. The growing number of investors in Indonesia reflects an increasing interest in capital market instruments, particularly in the consumer non-cyclicals sector, which is known for its stability and resilience to economic fluctuations. This study aims to: (1) compare the performance of stock price clustering in the consumer noncyclicals sector using two dissimilarity measures—Complexity Invariant Distance (CID) and Dynamic Time Warping (DTW), (2) identify the characteristics of stock price movements within each resulting cluster, and (3) forecast stock prices at the cluster level using the Triple Exponential Smoothing (Holt-Winter’s) method with both multiplicative and additive models. DTW was chosen for its ability to align time series with varying lengths or patterns, while CID was selected for its sensitivity to the complexity of time series shapes. The data used consist of weekly closing prices from 129 listed companies in the consumer non-cyclicals sector on the Indonesia Stock Exchange from January 1, 2019, to December 30, 2024. The results indicate that DTW with average linkage in hierarchical agglomerative clustering produces the best clustering outcome, with four optimal clusters and a silhouette coefficient of 0.75, indicating strong clustering quality. Forecasting using the Holt-Winter’s method yielded excellent predictive performance, with MAPE values below 10%: 4.62% for Cluster A, 3.24% for Cluster B, 3.06% for Cluster C, and 3.82% for Cluster D. These findings offer an efficient and accurate approach to understanding stock price movement patterns and support more informed investment decision-making in the consumer non-cyclicals sector. |
| URI: | http://repository.ipb.ac.id/handle/123456789/166412 |
| Appears in Collections: | UT - Statistics and Data Sciences |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G1401211026_03e3547fc7b647e5983802e5f5f89d02.pdf | Cover | 289.41 kB | Adobe PDF | View/Open |
| fulltext_G1401211026_d06b85e544984d1e9cdea4496a791b6d.pdf Restricted Access | Fulltext | 1.77 MB | Adobe PDF | View/Open |
| lampiran_G1401211026_afa3babf96254e078749e481293253db.pdf Restricted Access | Lampiran | 6.98 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.