BIAS REDUCTION IN KERNEL ESTIMATOR OF DENSITY FUNCTION IN BOUNDARY REGION

The properties of the classical kernel estimator of density function deteriorate when the support of density function is bounded. The use of classical form of kernel estimator causes the increase of the bias estimator, particularly in the so-called boundary region, close to end of support. It can also lead to undesirable situation where density function estimator has a different support than the density function. The paper presents selected bias reduction procedures, such as reflection method and its modification. An example is presented with an attempt to compare considered procedures.