Estimator selection for regression functions in exponential families with application to changepoint detection

12/25/2022
by   Juntong Chen, et al.
0

We observe n independent pairs of random variables (W_i, Y_i) for which the conditional distribution of Y_i given W_i=w_i belongs to a one-parameter exponential family with parameter γ^*(w_i)∈ℝ and our aim is to estimate the regression function γ^*. Our estimation strategy is as follows. We start with an arbitrary collection of piecewise constant candidate estimators based on our observations and by means of the same observations, we select an estimator among the collection. Our approach is agnostic to the dependencies of the candidate estimators with respect to the data and can therefore be unknown. From this point of view, our procedure contrasts with other alternative selection methods based on data splitting, cross validation, hold-out etc. To illustrate its theoretical performance, we establish a non-asymptotic risk bound for the selected estimator. We then explain how to apply our procedure to the changepoint detection problem in exponential families. The practical performance of the proposed algorithm is illustrated by a comparative simulation study under different scenarios and on two real datasets from the copy numbers of DNA and British coal disasters records.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
03/13/2022

Estimating a regression function in exponential families by model selection

Let X_1=(W_1,Y_1),…,X_n=(W_n,Y_n) be n pairs of independent random varia...
research
11/03/2020

Robust estimation of a regression function in exponential families

We observe n pairs X_1=(W_1,Y_1),…,X_n=(W_n,Y_n) of independent random v...
research
07/07/2023

Robust estimation for ergodic Markovian processes

We observe n possibly dependent random variables, the distribution of wh...
research
02/24/2021

Estimation and testing on independent not identically distributed observations based on Rényi's pseudodistances

In real life we often deal with independent but not identically distribu...
research
03/22/2023

Generalized Data Thinning Using Sufficient Statistics

Our goal is to develop a general strategy to decompose a random variable...
research
10/08/2020

Adaptive spectral density estimation by model selection under local differential privacy

We study spectral density estimation under local differential privacy. A...
research
06/16/2020

Decomposable Families of Itemsets

The problem of selecting a small, yet high quality subset of patterns fr...

Please sign up or login with your details

Forgot password? Click here to reset