Reconstruction of Conditional Expectations from Product Moments with Applications
Robert C. Hagwood
Regression calibration is a method for fitting models when the independent variable X is measured with error. In regression calibration, the noisy independent measurement W for X is calibrated with E[X | W] and the model is fitted using whatever method is appropriate using this replacement for X. An important component of this approach is using the data to find a good estimator of E[X | W]. In this paper, the function E[X | W = w] is estimated using the classical moment problem.
Reconstruction of Conditional Expectations from Product Moments with Applications, Journal of Computational and Applied Mathematics, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=908278
(Accessed May 29, 2023)