grid_jitter: Remove over-plotting by gridding point data intelligently

In geotheory/gridPoints: Functions for gridding point data

Remove over-plotting by gridding point data intelligently

Description

Data jittering reduces overplotting by adding small variances to values. grid_jitter removes it entirely by fitting points to a custom grid. This function applies Hungarian algorithm to match entire points set to the whole grid. (By contrast grid_jitter2 applies Hungarian with small area constraints.) This function works well for smaller grids and point sets, but will slow considerably with larger sets.

Usage

grid_jitter(x, y = NULL, nx = 50, ny = NULL, tol = 5,
  plotresults = TRUE, file = NULL, w = 20, h = 20)

Arguments

x	Numeric vector or 2 column matrix or data.frame of data points to plot
y	Numeric vector, y coordinates matching x (if a vector)
nx	Numeric, grid x/y dimensions
ny	Numeric
tol	Numeric, the maximum distance overplotted points are allowed to move to the nearest vacant grid cell
plotresults	Logical, output plots to illustrate point displacements and cell-reallocations
file	String, if not NULL the result of plotresults will save to filename instead of rendering in R
w	Numeric, width/height (inches) of plotresults if output to file (PDF)
h	Numeric

Value

A 2 column matrix of grid-jittered point coordinates

Examples

# 1. normal distribution
d1 = data.frame(x = rnorm(300), y = rnorm(300))
d1g = grid_jitter(d1, nx=50, ny=50, tol=5)

# 2. faithful dataset
d2 = faithful
d2g = grid_jitter(d2, nx=50, ny=50, tol=3)

# 3. mpg example
d3 = ggplot2::mpg[,c(3,8)]
d3g = grid_jitter(d3, nx=100, ny=100, tol=3)

# 4. US States
d4 = data.frame(x=state.center$x, y=state.center$y, id=state.abb)
d4j = grid_jitter(d4[,1:2], nx=10, ny=8, tol=3, plotresults=TRUE)
cols = sample(colorRampPalette(c('darkblue','blue','lightblue'))(50))
par(mai=rep(.6,4), mfrow=c(1,2))
plot(d4[,1:2], pch=15, cex=2.5, col=cols, asp=2.5, xlab=NA, ylab=NA)
text(d4[,1:2], labels=state.abb, col='white', cex=.7)
plot(d4j, pch=15, cex=2.5, col=cols, asp=2.5, xlab=NA, ylab=NA)
text(d4j, labels=d4$id, col='white', cex=.7)

# 5. to illuste cell reallocation 
d5 = data.frame(x = c(1,rep(50, 398), 100), y = c(1,rep(50, 398), 100))
d5g = grid_jitter(d5, nx=30, ny=30, tol=15)

geotheory/gridPoints documentation built on March 25, 2022, 5 p.m.