Title: On a regression model with constraints in Hilbert spaces
Authors: Ana Colubi - Kings College London (United Kingdom)
Gil Gonzalez-Rodriguez - University of Oviedo (Spain)
Marta Garcia Barzana - Universidad de Oviedo (Spain) [presenting]
Abstract: The least-squares estimation of linear regression models involves an optimization problem that may be subject to a certain group of constraints. The well-known constrained least-squares approach assumes that the number of inequality linear constrains is fixed. This framework is extended by removing such an assumption. Thus, the number of constrains can vary depending on the sample size. This problem has been addressed in the context of linear regression with interval data. However, the goal is to extend the problem to the abstract case of regression models in Hilbert spaces, which accommodates as well more complex data, such as functional data. An estimator is proposed and a case-based example is presented.