ABSTRACT
Densi ty est imation and nonparametric regression are the most fundamental and familiar problems where kernel smoothing techniques provide an effective solut ion. However, the principles of kernel smoothing can be generalised and adapted to overcome several more complicated problems. These can broadly be d iv ided into: • si tuations where the da ta do not conform w i t h the usual ran-
dom sample assumptions, such as those where the da ta have dependencies, or are observed wi th error;
• est imation of other functions. Examples include hazard rates, spectral densities and intensity functions.
