Functional regression models with complex structure

Functional data analysis is a set of methodologies suitable for the analysis of high-dimensional measurements, such as curves or surfaces, which consider data not as a sequence of single measurements taken one after another, but as whole functional entities. Regression models are considered to be functional if the explanatory variable, the dependent variable, or both the explanatory and dependent variables can be treated as functions. The aim of the dissertation is the development of suitable statistical methods and algorithms mainly focused on statistical modelling when the random variables have a complex variation and correlation structure, there are restrictions on regression parameters, or observations are incomplete. The emphasis will be given both on theoretical aspects concerning estimation, uncertainty and statistical inference, as well as practical implementation and computational feasibility.