R: Partial distance matrix (focus on closest)

partialDist {wrMisc}

R Documentation

Partial distance matrix (focus on closest)

Description

partialDist calculates distance matrix like dist for 1- or 2-dim data, but only partially, ie only cases of small distances. This function was made for treating very large data-sets where only very close distances to a given point need to be found, it allows to overcome memory-problems with larger data (and faster execution with > 50 rows of 'dat').

Usage

partialDist(
  dat,
  groups,
  overLap = TRUE,
  method = "euclidean",
  silent = FALSE,
  debug = FALSE,
  callFrom = NULL
)

Arguments

`dat`	(matrix of numeric values) main input
`groups`	(factor) to split using `cut` or specific custom grouping (length of dat)
`overLap`	(logical) if TRUE make groups overlapping by 1 value (ie maintain some context-information)
`method`	'character' name of method passed to `dist`
`silent`	(logical) suppress messages
`debug`	(logical) display additional messages for debugging
`callFrom`	(character) allow easier tracking of message(s) produced

Value

This function returns a matrix with partial distances (not of class 'dist' object)

Examples

set.seed(2016); mat3 <- matrix(runif(300),nr=30)
round(dist(mat3), 1)
round(partialDist(mat3, gr=3), 1)