Non-parametric Poisson regression from independent and weakly dependent observations by model selection

Abstract

We consider the non-parametric Poisson regression problem where the integer valued response Y is the realization of a Poisson random variable with parameter λ(X). The aim is to estimate the functional parameter λ from independent or weakly dependent observations (X1, Y1), . . . , (Xn, Yn) in a random design framework.First we determine upper risk bounds for projection estimators on finite dimensional subspaces under mild conditions. In the case of Sobolev ellipsoids the obtained rates of convergence turn out to be optimal.The main part of the paper is devoted to the construction of adaptive projection estimators of λ via model selection. We proceed in two steps: first, we assume that an upper bound for λ ∞ is known. Under this assumption, we construct an adaptive estimator whose dimension parameter is defined as the minimizer of a penalized contrast criterion. Second, we replace the known upper bound on λ ∞ by an appropriate plug-in estimator of λ ∞. The resulting adaptive estimator is shown to attain the minimax optimal rate up to an additional logarithmic factor both in the independent and the weakly dependent setup. Appropriate concentration inequalities for Poisson point processes turn out to be an important ingredient of the proofs.We illustrate our theoretical findings by a short simulation study and conclude by indicating directions of future research.

Adaptive spectral density estimation by model selection under local differential privacy

“…Theoretical results can be proved for this more realistic estimator as well. We do not realize this here, and refer the interested reader to the papers [Com01] and [Kro19] where this idea has been put into practise. The resulting fully-adaptive estimator can be shown to satisfy an oracle inequality as in the case of known f ∞ under mild assumptions.…”

Gaussianization Machines for Non-Gaussian Function Estimation Models

“…Nonparametric Poisson regression is of significant interest in its own right. Regression with count data arises in a range of applications, see, for example, Ver Hoef and Boveng (2007), Winkelmann (2003), Berk and MacDonald (2008), Kroll (2019). The Poisson regression model is one of the most natural approaches to count data regression.…”

Law of the Iterated Logarithm and Model Selection Consistency for GLMs with Independent and Dependent Responses

“…Some work has paid attention to linear models with correlated errors, see Fan et al (2016). However, in contrast to linear models with dependent responses, the GLMs emphasizing on dependent errors or dependent responses have only been scarcely investigated, see Kroll (2019). This section focuses on the GLMs endowed with weakly dependent responses, i.e.…”