Abstract
Statistical inference can be over optimistic and even misleading based on a selected model due to the uncertainty of the model selection procedure, especially in the highdimensional data analysis. In this article, we propose a bootstrap-based tilted correlation screening learning (TCSL) algorithm to alleviate this uncertainty. The algorithm is inspired by the recently proposed variable selection method, TCS algorithm, which screens variables via tilted correlation. Our algorithm can reduce the prediction error and make the interpretation more reliable. The other gain of our algorithm is the reduced computational cost compared with the TCS algorithm when the dimension is large. Extensive simulation examples and the analysis of one real dataset are conducted to exhibit the good performance of our algorithm. Supplementary materials for this article are available online.
Original language | English |
---|---|
Pages (from-to) | 478-496 |
Number of pages | 19 |
Journal | Journal of Computational and Graphical Statistics |
Volume | 23 |
Issue number | 2 |
DOIs | |
Publication status | Published - 1 Jan 2014 |
Externally published | Yes |
Keywords
- Bootstrap
- Model averaging
- TCS algorithm
- Variable selection
ASJC Scopus subject areas
- Discrete Mathematics and Combinatorics
- Statistics and Probability
- Statistics, Probability and Uncertainty