epsilonCompute: Compute default diffusion map epsilon.

In diffusionMap: Diffusion Map

Description

Uses the pair-wise distances to estimate a diffusion map epsilon value by the median p*n-th nearest neighbor

Usage

epsilonCompute(D, p = 0.01)

Arguments

D	n-by-n pairwise distance matrix for a data set with n points, or alternatively output from the dist() function
p	distances to p*n-th nearest neighbor are used. Default value is .01

Details

Function is used as the default value in diffuse(). For inference problems, it is advised that the results be optimized over epsilon.

Value

epsilon

value of epsilon to be used in diffusion map

See Also

diffuse()

Examples

data(annulus)
D = dist(annulus) # use Euclidean distance
epsilonCompute(D,.005)
epsilonCompute(D,.01)
epsilonCompute(D,.05)
epsilonCompute(D,.1)

Example output

[1] 0.05670553
[1] 0.1223964
[1] 1.45663
[1] 3.585611

diffusionMap documentation built on Sept. 11, 2019, 1:04 a.m.