Estimators of variance function in nonparametric regression

The thesis studies variance function estimation in nonparametric regression model. It focuses on local polynomial estimators particularly. Exact expressions of conditional variance function estimator bias and covariance are derived and important asymptotical aproximations of these characteristics are also provided. Further the EBBS method for bandwidth selection and Dette's homoscedasticity test are described. Results of Prague Klementinum data processing are presented at the end of the thesis