The lasso for high dimensional regression with a possible change point
- Publication Type:
- Journal Article
- Citation:
- Journal of the Royal Statistical Society. Series B: Statistical Methodology, 2016, 78 (1), pp. 193 - 210
- Issue Date:
- 2016-01-01
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| 1209.4875v4.pdf | Accepted Manuscript Version | 831.82 kB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
© 2015 The Authors Journal of the Royal Statistical Society: Series B (Statistics in Society) Published by John Wiley & Sons Ltd on behalf of the Royal Statistical Society. We consider a high dimensional regression model with a possible change point due to a covariate threshold and develop the lasso estimator of regression coefficients as well as the threshold parameter. Our lasso estimator not only selects covariates but also selects a model between linear and threshold regression models. Under a sparsity assumption, we derive non-asymptotic oracle inequalities for both the prediction risk and the (Formula presented.) -estimation loss for regression coefficients. Since the lasso estimator selects variables simultaneously, we show that oracle inequalities can be established without pretesting the existence of the threshold effect. Furthermore, we establish conditions under which the estimation error of the unknown threshold parameter can be bounded by a factor that is nearly (Formula presented.) even when the number of regressors can be much larger than the sample size n. We illustrate the usefulness of our proposed estimation method via Monte Carlo simulations and an application to real data.
Please use this identifier to cite or link to this item: