Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/170126| Title: | Penggerombolan Spasio-Temporal Menggunakan K-Medoids dengan Jarak Euclidean Terbobot pada Harga Telur di Indonesia |
| Other Titles: | |
| Authors: | Alamudi, Aam Anisa, Rahma BREGINA, IGNACIA MANUELA |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Harga telur ayam di Indonesia mengalami fluktuasi yang cukup besar selama periode 2022 hingga 2024, yang menunjukkan adanya perbedaan pola harga antar provinsi. Oleh karena wilayah Indonesia sangat luas, variasi harga terjadi dari waktu ke waktu dan antar lokasi sehingga dibutuhkan pendekatan yang mampu menangkap pola tersebut secara menyeluruh. Penelitian ini bertujuan untuk mengidentifikasi pola spasio-temporal harga telur, mengelompokkan provinsi berdasarkan kemiripan pola harga, serta mengevaluasi metode yang digunakan. Metode yang digunakan melibatkan penghitungan jarak temporal dengan Dynamic Time Warping (DTW) dan jarak spasial menggunakan matriks biner dari K-Nearest Neighbors (KNN) berbasis Great-Circle Distance (GCD), yang digabungkan melalui Weighted Euclidean Distance (WED). Hasil penelitian menunjukkan kombinasi bobot terbaik adalah 0,6 untuk spasial dan 0,4 untuk temporal, dengan nilai Silhouette sebesar 0,6438 dan cost sebesar 76.399,2. Jumlah tetangga optimal ditentukan sebanyak 2 provinsi, dengan nilai indeks Moran sebesar 0,6759. Proses pengelompokan menghasilkan empat gerombol dengan nilai Silhouette = 0,5881 dan DBI = 0,3349, serta dikonfirmasi oleh plot Elbow. Setiap gerombol dianalisis berdasarkan tingkat kestabilan harga antar wilayah selama periode observasi. The price of chicken eggs in Indonesia has experienced significant fluctuations during the period from 2022 to 2024, indicating differences in price patterns between provinces. Due to Indonesia's vast territory, price variations occur over time and between locations, necessitating an approach that can capture these patterns comprehensively. This study aims to identify spatiotemporal price patterns for eggs, group provinces based on similarity in price patterns, and evaluate the methods used. The methods employed involve calculating temporal distance using Dynamic Time Warping (DTW) and spatial distance using a binary matrix from K-Nearest Neighbors (KNN) based on Great-Circle Distance (GCD), combined through Weighted Euclidean Distance (WED). The results indicate that the optimal weight combination is 0.6 for spatial and 0.4 for temporal, with a Silhouette value of 0.6438 and a cost of 76,399.2. The optimal number of neighbors is determined to be 2 provinces, with a Moran index value of 0.6759. The clustering process resulted in four clusters with a Silhouette value of 0.5881 and a DBI of 0.3349, confirmed by the Elbow plot. Each cluster was analyzed based on the stability of prices between regions during the observation period. |
| URI: | http://repository.ipb.ac.id/handle/123456789/170126 |
| Appears in Collections: | UT - Statistics and Data Sciences |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G1401211072_8bd43dc7bc5245daa86237312bab8e5b.pdf | Cover | 1.78 MB | Adobe PDF | View/Open |
| fulltext_G1401211072_cade8a6455d24229b8bdb57b1ec66588.pdf Restricted Access | Fulltext | 3.44 MB | Adobe PDF | View/Open |
| lampiran_G1401211072_4d614be1ab814b73ab51af4d2fd49a55.pdf Restricted Access | Lampiran | 747.78 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.