|
キョウ コウキ
Kyo Koki
姜 興起 所属 デジタルトランスフォーメーション（DX）推進センター 職種 教授 |
|
| 言語種別 | 英語 |
| 発行・発表の年月 | 2023/11/15 |
| 形態種別 | 研究論文（学術雑誌） |
| 査読 | 査読あり |
| 標題 | A moving linear model approach for extracting cyclical variation from time series data |
| 執筆形態 | 共著 |
| 掲載誌名 | Journal of Business Cycle Research |
| 掲載区分 | 国外 |
| 巻・号・頁 | 19(3),pp.373-397 |
| 総ページ数 | 25 |
| 担当区分 | 筆頭著者 |
| 著者・共著者 | Koki Kyo and Genshiro Kitagawa |
| 概要 | We introduce a method to decompose time series data into various components, such as constrained and cyclical components, using a moving linear modeling approach and state space representation. The critical parameter is the width of the time interval, estimated via maximum likelihood. Importantly, a local linear model suffices for the constrained component, rather than a strict one for the remainder. Iteratively applying our method decomposes time series into multiple components. Additionally, we outline a procedure to transform these components into uncorrelated ones through principal component analysis. This approach proves useful in analyzing business cycles, as evidenced by its application to monthly data from Japan. |