Density Estimation for Exploring Data

Abstract

Abstract The concept of a probability density function is a central idea in statistics. Its role in statistical modelling is to encapsulate the pattern of random variation in the data which is not explained by the other structural terms of a model. In many settings this role, while important, is a secondary one, with the principal focus resting on the nature of covariate or other effects. However, there are also situations where the detailed shape of the underlying density function is itself of primary interest. In this chapter, a number of examples of this are given and the ideas behind the construction of a smooth estimate of a density function are introduced. Some of the main issues associated with using these estimates are raised.