AEdist.meas: Asymptotic Expected Distance in Measure (AEDM)

In AlessandroCasa/BsMc: Bandwidth selection for modal clustering

Description

Compute the Asymptotic Expected Distance in Measure (AEDM) between a kernel estimator-induced partition and the population one defined by a K-component normal mixture density over a grid of bandwidths.

Usage

AEdist.meas(n = 100, hmin = 0.05, hmax = 1, byh = hmin, mus = 0,
  sigmas = 1, props = 1, plot = TRUE)

Arguments

n	sample size of the simulated Monte Carlo samples.
hmin	lower value for the grid of bandwidths.
hmax	upper value for the grid of bandwidths.
byh	increment of the grid sequence
mus	vector of means of the mixture components.
sigmas	vector of standard deviations of the mixture components.
props	vector of mixing proportions of the mixture components.
plot	if TRUE the curve of the AEDM as a function of bandwidth values in the grid is displayed along with the single DM for each Monte Carlo sample.

Value

the value of the EDM for each value in the grid.

References

Casa A., Chacón, J.E. and Menardi, G. (2019). Modal clustering asymptotics with applications to bandwidth selection (https://arxiv.org/pdf/1901.07300.pdf).

AlessandroCasa/BsMc documentation built on Oct. 30, 2019, 4:49 a.m.