dpCluster: dpCluster
In evanwang1990/dpCluster: Clustering by fast search and find of density peaks

Description Usage Arguments

dpCluster

1
2
3

dpCluster(x, percent = 0.01, thres.rho = NULL, thres.delta = NULL,
  similarity = c("euclidean", "SNN"), method = c("gaussian", "withinDc",
  "neighbors"), threads = 4, halo.detection = TRUE)

`x`	data.frame or matrix
`percent`	method 'gaussian' and 'withinDc' use the nearest 'percent' of points to estimate 'dc'; method 'neighbors' use the nearest 'percent' of points to estimate density directly. percent is suggested between 0.01 and 0.02
`thres.rho`	threshold of rho to detect peaks, which is often get from "Decision Graph".
`thres.delta`	threshold of delta to detect peaks, which is often get from "Decision Graph".
`method`	there are three methods to estimate density. 'gaussian' : gaussian kernel density estimation $e^(\frac-12(\frac\delta(x_i, x_j)2dc)^2)$ 'withinDc' : the density of $x_i$ is defined as the number of points within dc distance of $x_i$ 'neighbors' : the density of $x_i$ is defined as the reciprocal of the mean distance of $x_i$'s neighbors
`threads`	number of threads to use for task scheduling
`halo.detection`	logical, whether to remove the potential noise point with low density