| Item type |
デフォルトアイテムタイプ（フル）その２(1) |
| 公開日 |
2025-02-12 |
| タイトル |
|
|
タイトル |
Expected Shortfall Regression for High-Dimensional Additive Models |
|
言語 |
en |
| 作成者 |
本田, 敏雄
PENG, Po-Hsiang
|
| アクセス権 |
|
|
アクセス権 |
open access |
|
アクセス権URI |
http://purl.org/coar/access_right/c_abf2 |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
expected shortfall |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
quantile regression |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
group Lasso |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
group SCAD |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
B-spline basis |
| 主題 |
|
|
言語 |
en |
|
主題Scheme |
Other |
|
主題 |
additive models |
| 出版者 |
|
|
出版者 |
Graduate School of Economics, Hitotsubashi University |
| 日付 |
|
|
日付 |
2025-02 |
|
日付タイプ |
Issued |
| 言語 |
|
|
言語 |
eng |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_18gh |
|
資源タイプ |
technical report |
| 出版タイプ |
|
|
出版タイプ |
VoR |
|
出版タイプResource |
http://purl.org/coar/version/c_970fb48d4fbd8a85 |
| 関連情報 |
|
|
関連タイプ |
isPartOf |
|
|
関連名称 |
Discussion papers ; No. 2025-01 |
| 助成情報 |
|
|
|
助成機関識別子タイプ |
Crossref Funder |
|
|
助成機関識別子 |
https://doi.org/10.13039/501100001691 |
|
|
助成機関名 |
日本学術振興会 |
|
|
言語 |
ja |
|
|
助成機関名 |
Japan Society for the Promotion of Science |
|
|
言語 |
en |
|
|
研究課題番号URI |
https://kaken.nii.ac.jp/ja/grant/KAKENHI-PROJECT-24K14850/ |
|
|
研究課題番号 |
24K14850 |
|
|
研究課題名 |
構造を持つノンパラメトリック回帰モデルによる超高次元データ解析に関する研究 |
|
|
言語 |
ja |
| ページ数 |
|
|
ページ数 |
37 |
| Sponsorship |
|
|
値 |
This research is financially supported by JSPS KAKENHI Grant Number JP 24K14850 (HONDA) and Taiwan NSTC Grant Number 112-2122-M-007-001-MY3 (PENG). |
| 抄録（第三者提供不可） |
|
|
値 |
The expected shortfall (ES) regression can be a powerful and useful tool to analyze the relation between the response variable and the covariates through the conditional mean. As is well-known, there is no single loss function for expected shortfall estimation and there is a suitable loss function for joint estimation of quantile and expected shortfall. In addition to them, recently a very useful two-step procedure for ES regression was proposed : carry out quantile regression and then estimate the ES regression model by applying the least squares method. This procedure is successful due to the Neyman orthogonality. Then high dimensional linear regression models was considered based on the the findings. By exploiting those results, we assume additive models for both quantile and expected shortfall in the high-dimensional setting and consider the group Lasso and SCAD estimators. We establish the oracle inequality and the oracle property for them. Our theoretical results also imply that quantile estimation does not affect ES estimation asymptotically. We also present numerical results that demonstrate satisfactory performance in model selection, estimation accuracy, and prediction error for a moderate sample size together with an empirical study. |
070econDP25-01.pdf (491.2 KB)
