40
8

Inference for local parameters in convexity constrained models

Abstract

We consider the problem of inference for local parameters of a convex regression function f0:[0,1]Rf_0: [0,1] \to \mathbb{R} based on observations from a standard nonparametric regression model, using the convex least squares estimator (LSE) f^n\widehat{f}_n. For x0(0,1)x_0 \in (0,1), the local parameters include the pointwise function value f0(x0)f_0(x_0), the pointwise derivative f0(x0)f_0'(x_0), and the anti-mode (i.e., the smallest minimizer) of f0f_0. The existing limiting distribution of the estimation error (f^n(x0)f0(x0),f^n(x0)f0(x0))(\widehat{f}_n(x_0) - f_0(x_0), \widehat{f}_n'(x_0) - f_0'(x_0) ) depends on the unknown second derivative f0(x0)f_0''(x_0), and is therefore not directly applicable for inference. To circumvent this impasse, we show that the following locally normalized errors (LNEs) enjoy pivotal limiting behavior: Let [u^(x0),v^(x0)][\widehat{u}(x_0), \widehat{v}(x_0)] be the maximal interval containing x0x_0 where f^n\widehat{f}_n is linear. Then, under standard conditions, \binom{ \sqrt{n(\widehat{v}(x_0)-\widehat{u}(x_0))}(\widehat{f}_n(x_0)-f_0(x_0)) }{ \sqrt{n(\widehat{v}(x_0)-\widehat{u}(x_0))^3}(\widehat{f}_n'(x_0)-f_0'(x_0))} \rightsquigarrow \sigma \cdot \binom{\mathbb{L}^{(0)}_2}{\mathbb{L}^{(1)}_2}, where nn is the sample size, σ\sigma is the standard deviation of the errors, and L2(0),L2(1)\mathbb{L}^{(0)}_2, \mathbb{L}^{(1)}_2 are universal random variables. This asymptotically pivotal LNE theory instantly yields a simple tuning-free procedure for constructing CIs with asymptotically exact coverage and optimal length for f0(x0)f_0(x_0) and f0(x0)f_0'(x_0). We also construct an asymptotically pivotal LNE for the anti-mode of f0f_0, and its limiting distribution does not even depend on σ\sigma. These asymptotically pivotal LNE theories are further extended to other convexity/concavity constrained models (e.g., log-concave density estimation) for which a limit distribution theory is available for problem-specific estimators.

View on arXiv
Comments on this paper