dbscan: Density Based Clustering

In Armadilloa16/dipps: Tools for Bruker Peaklist Format MSI Data

Description

This density based clustering is a univariate optimisation of the DBSCAN* algorithm as in section 3 of Campello et al. (2013).

Usage

dbscan(x, eps = 0.05, mnpts = 100, pp = FALSE)

Arguments

x	A numeric vector of values to be clustered
eps	A positive number representing the window radius in which to group points / measure density.
mnpts	A positive number representing the minimum density for a point to be included – the minimum number of points within +/- eps.
pp	A logical value indicating whether progress should be printed to the console.

Details

Note that a mnpts of 1 will produce groups that correspond to the equivalence classes of the relation “are within eps of each other”. This can be useful for low-density or low-noise data.

This is included in this package due to its usefulness in clustering peaks by mass, see example below.

See Also

combine_peaklists

Section 3 of Campello, Ricardo JGB, Davoud Moulavi, and Joerg Sander. "Density-based clustering based on hierarchical density estimates." In Advances in Knowledge Discovery and Data Mining, pp. 160-172. Springer Berlin Heidelberg, 2013.

Examples

i.path = system.file("extdata", "test1", package = "dipps")
n.emp  = combine_peaklists(i.path)
o.name = basename(i.path)
df.peak = load_peaklist(o.name)
df.peak$group = dbscan(df.peak$m.z)

Armadilloa16/dipps documentation built on May 5, 2019, 7:06 a.m.