nnhclu

Nearest-neighbour hierarchical clustering

R

nnhclu(
  x,
  y = NULL,
  area = NULL,
  fixd = NULL,
  p = 0.5,
  minpts = 10,
  minhi = minpts,
  maxord = 3,
  k = 1,
  c = 1,
  count = NULL,
  lonlat = FALSE
)

Groups points that lie closer together than a threshold distance, then groups the groups, order by order (after Johnson 1967, as used for crime hot spots). At the first order two points are linked when their distance is below the threshold: fixd, or the mean nearest-neighbour distance expected under complete spatial randomness in an area area plus z standard errors, 0.5 sqrt(A / m) + z sqrt((4 - pi) A / (4 pi)) / m for m units, with z = qnorm(p) (p = 0.5 gives the expectation itself, smaller p a stricter threshold). Points with a linked neighbour are taken in decreasing number of links; each one not yet clustered takes every unclustered point linked to it; groups of at least minpts points are kept. Their centres of minimum distance become the units of the next order, with the threshold recomputed for their number, while at least four remain, up to maxord orders. The iterative reassignment some descriptions add after the first pass is not done: its rule is not stated unambiguously.

Parameters

x, y

The points: x and y numeric vectors of equal length; or x a matrix or data frame with columns named x and y (or X and Y, as sf::st_coordinates() names them, or lon and lat with lonlat = TRUE) and y NULL; or x an sf or sfc object of POINT geometries, whose coordinate reference system must agree with lonlat. An unnamed two-column matrix is read as x then y; other columns are never taken by position. Vectors given separately (x, y, and w or t where a function takes them) are matched by name when every one is named and no name repeats, as elsewhere in the package; otherwise by position. No missing or infinite values.

area

The study area for the random-expectation threshold, in the square of the coordinates’ unit (square metres when lonlat = TRUE). Give area or fixd, not both.

fixd

A fixed threshold distance instead, positive.

p

The one-tailed probability that a pair is as close by chance: one number strictly between 0 and 1.

minpts, minhi

The fewest points (order 1) and member units (higher orders) a cluster keeps: whole numbers from 2.

maxord

The highest order: one whole number from 1 to 20.

k

The divisor offset: 0, 1 or 2; the total weight must exceed it.

c

The multiplier of each cluster’s standard deviational ellipse (see sdell()).

count

NULL for points, or for zones: one whole count from 0 per zone, the points x, y being the zones’ centroids. A zone is taken as count incidents at its centroid, which these hot-spot rules treat exactly as coincident points. Zones with count 0 get NA; the random-expectation threshold counts incidents, not zones.

lonlat

FALSE (the default) when x and y are planar coordinates in one length unit; TRUE when they are longitude and latitude in degrees, which are then projected to metres by eqproj() about their own centre, and centres are returned in degrees. It is never guessed.

Returns

A list. cl, an integer matrix with one row per point (input order) and one column per order: the cluster (a row of clusters) the point belongs to at that order, or NA. clusters, a data frame with one row per cluster: order, points, units (member units), mx, my (mean centre), wx, wy (centre of minimum distance; both in degrees when lonlat = TRUE), a, b, angle (the ellipse of the member units, NaN when too few for the divisor) and thresh (the threshold distance used).

References

Johnson, S. C. (1967). Hierarchical clustering schemes. Psychometrika, 32(3), 241-254. doi:10.1007/BF02289588

Eck, J. E., Chainey, S., Cameron, J. G., Leitner, M. and Wilson, R. E. (2005). Mapping Crime: Understanding Hot Spots. National Institute of Justice, NCJ 209393.

Examples

set.seed(1)
x <- c(rnorm(40, 0, 0.3), rnorm(40, 6, 0.3), runif(40, -5, 10))
y <- c(rnorm(40, 0, 0.3), rnorm(40, 6, 0.3), runif(40, -5, 10))
nnhclu(x, y, area = 225, p = 0.05, minpts = 5)$clusters

Generated from the R help page: nnhclu in the R reference.