English

Sparse Quantile Regression

Methodology 2023-03-30 v4 Econometrics

Abstract

We consider both 0\ell _{0}-penalized and 0\ell _{0}-constrained quantile regression estimators. For the 0\ell _{0}-penalized estimator, we derive an exponential inequality on the tail probability of excess quantile prediction risk and apply it to obtain non-asymptotic upper bounds on the mean-square parameter and regression function estimation errors. We also derive analogous results for the 0\ell _{0}-constrained estimator. The resulting rates of convergence are nearly minimax-optimal and the same as those for 1\ell _{1}-penalized and non-convex penalized estimators. Further, we characterize expected Hamming loss for the 0\ell _{0}-penalized estimator. We implement the proposed procedure via mixed integer linear programming and also a more scalable first-order approximation algorithm. We illustrate the finite-sample performance of our approach in Monte Carlo experiments and its usefulness in a real data application concerning conformal prediction of infant birth weights (with n103n\approx 10^{3} and up to p>103p>10^{3}). In sum, our 0\ell _{0}-based method produces a much sparser estimator than the 1\ell _{1}-penalized and non-convex penalized approaches without compromising precision.

Keywords

Cite

@article{arxiv.2006.11201,
  title  = {Sparse Quantile Regression},
  author = {Le-Yu Chen and Sokbae Lee},
  journal= {arXiv preprint arXiv:2006.11201},
  year   = {2023}
}

Comments

51 pages, 3 figures, 3 tables