colourvalues

What does it do?

It maps viridis colours (by default) to values, and quickly!

Note It does not perform a 1-to-1 mapping of a palette to values. It interpolates the colours from a given palette.

Why did you build it?

I’m aware there are other methods for mapping colours to values. And which do it quick too. But I can never remember them, and I find the interfaces a bit cumbersome. For example, scales::col_numeric(palette = viridisLite::viridis(5), domain = range(1:5))(1:5).

I wanted one function which will work on one argument.

colour_values(1:5)
# [1] "#440154FF" "#3B528BFF" "#21908CFF" "#5DC963FF" "#FDE725FF"
colour_values(letters[1:5])
# [1] "#440154FF" "#3B528BFF" "#21908CFF" "#5DC963FF" "#FDE725FF"

I also want it available at the src (C/C++) level for linking to other packages.

Why do you spell colour with a ‘u’?

Because it’s correct, and R tells us to

For consistency, aim to use British (rather than American) spelling

But don’t worry, color_values(1:5) works as well

How do I install it?

From CRAN

install.packages("colourvalues")

Or install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("SymbolixAU/colourvalues")

How can I make use of it in my package?

Rcpp

All functions are written in Rcpp. I have exposed some of them in header files so you can “link to” them in your package.

For example, the LinkingTo section in DESCRIPTION will look something like

LinkingTo: 
    Rcpp,
    colourvalues

And in a c++ source file so you can #include the API header

#include "colourvalues/api.hpp"
// [[Rcpp::depends(colourvalues)]]

And call

// return hex colours
colourvalues::api::colour_values_hex()

// return RGP matrix
colourvalues::api::colour_values_rgb()

If you’re not using Rcpp, just Import this package like you would any other.

Do you have any examples?

Of course!

256 numbers mapped to a colour

bar_plot <- function(df) {
  barplot( height = df[["a"]], col = df[["col"]], border = NA, space = 0, yaxt = 'n')
}
df <- data.frame(a = 10, x = 1:256)
df$col <- colour_values(df$x, palette = "viridis")
bar_plot( df )

5000 numbers on a non-linear scale

df <- data.frame(a = 10, x = c((1:5000)**3))
df$col <- colour_values(df$x, palette = "viridis")
bar_plot( df )

1000 random numbers

df <- data.frame(a = 10, x = rnorm(n = 1000))
df$col <- colour_values(df$x, palette = "inferno")
bar_plot( df )

Eurgh!

df <- df[with(df, order(x)), ]
bar_plot( df )

That’s better!

Are there only viridis palettes?

No, you can chose one from

colour_palettes()
#  [1] "viridis"        "cividis"        "magma"          "inferno"       
#  [5] "plasma"         "ylorrd"         "ylorbr"         "ylgnbu"        
#  [9] "ylgn"           "reds"           "rdpu"           "purples"       
# [13] "purd"           "pubugn"         "pubu"           "orrd"          
# [17] "oranges"        "greys"          "greens"         "gnbu"          
# [21] "bupu"           "bugn"           "blues"          "spectral"      
# [25] "rdylgn"         "rdylbu"         "rdgy"           "rdbu"          
# [29] "puor"           "prgn"           "piyg"           "brbg"          
# [33] "terrain"        "topo"           "heat"           "cm"            
# [37] "rainbow"        "terrain_hcl"    "heat_hcl"       "sequential_hcl"
# [41] "rainbow_hcl"    "diverge_hcl"    "diverge_hsv"    "ygobb"         
# [45] "matlab_like2"   "matlab_like"    "magenta2green"  "cyan2yellow"   
# [49] "blue2yellow"    "green2red"      "blue2green"     "blue2red"

And you can use show_colours() to view them all. Here’s what some of them look like

show_colours( colours = colour_palettes(c("viridis", "colorspace")))

Do I have to use the in-built palettes?

No, you can use your own specified as a matrix of red, green and blue columns in the range [0,255]

n <- 100
m <- grDevices::colorRamp(c("red", "green"))( (1:n)/n )
df <- data.frame(a = 10, x = 1:n)
df$col <- colour_values(df$x, palette = m)
bar_plot( df )

Do you support ‘alpha’ values

Yep. Either supply a single alpha value for all the colours

## single alpha value for all colours
df <- data.frame(a = 10, x = 1:255)
df$col <- colour_values(df$x, alpha = 50)
bar_plot( df )

Or use a vector of values the same length as x

df <- data.frame(a = 10, x = 1:300, y = rep(c(1:50, 50:1), 3) )
df$col <- colour_values(df$x, alpha = df$y)
bar_plot( df )

Or include the alpha value as a 4th column in the palette matrix

n <- 100
m <- grDevices::colorRamp(c("red", "green"))( (1:n)/n )
## alpha values
m <- cbind(m, seq(0, 255, length.out = 100))
df <- data.frame(a = 10, x = 1:n)
df$col <- colour_values(df$x, palette = m)
bar_plot( df )

Some of my plotting functions don’t support alpha, can I exclude it?

Yep. Set include_alpha = FALSE

colour_values(1:5, include_alpha = F)
# [1] "#440154" "#3B528B" "#21908C" "#5DC963" "#FDE725"
colour_values_rgb(1:5, include_alpha = F)
#      [,1] [,2] [,3]
# [1,]   68    1   84
# [2,]   59   82  139
# [3,]   33  144  140
# [4,]   93  201   99
# [5,]  253  231   37

Can I get a summary of colours to use in a legend?

Yes, for numeric values use the n_summaries argument to specify the number of summary values you’d like

colour_values(1:10, n_summaries = 3)
# $colours
#  [1] "#440154FF" "#482878FF" "#3E4A89FF" "#31688EFF" "#26838EFF"
#  [6] "#1F9D89FF" "#35B779FF" "#6CCE59FF" "#B4DD2CFF" "#FDE725FF"
# 
# $summary_values
# [1] "1.00"  "5.50"  "10.00"
# 
# $summary_colours
# [1] "#440154FF" "#21908CFF" "#FDE725FF"

You can also specify the number of digits you’d like returned in the summary

colour_values(rnorm(n = 10), n_summaries = 3, digits = 2)
# $colours
#  [1] "#FDE725FF" "#3F4888FF" "#38578CFF" "#3C508BFF" "#2B768EFF"
#  [6] "#2A778EFF" "#463580FF" "#440154FF" "#287C8EFF" "#2A778EFF"
# 
# $summary_values
# [1] "-1.78" "0.53"  "2.84" 
# 
# $summary_colours
# [1] "#440154FF" "#21908CFF" "#FDE725FF"

You can also use format = FALSE if you don’t want the summary values formatted.

dte <- seq(as.Date("2018-01-01"), as.Date("2018-02-01"), by = 1)
colour_values(dte, n_summaries = 3)
# $colours
#  [1] "#440154FF" "#470D60FF" "#48196BFF" "#482474FF" "#472E7CFF"
#  [6] "#453882FF" "#414286FF" "#3E4B8AFF" "#3A548CFF" "#365D8DFF"
# [11] "#32658EFF" "#2E6D8EFF" "#2B758EFF" "#287D8EFF" "#25858EFF"
# [16] "#228C8DFF" "#20948CFF" "#1E9C89FF" "#20A386FF" "#25AB82FF"
# [21] "#2DB27DFF" "#39BA76FF" "#48C16EFF" "#58C765FF" "#6ACD5BFF"
# [26] "#7ED34FFF" "#92D742FF" "#A8DB34FF" "#BEDF26FF" "#D4E21BFF"
# [31] "#E9E41AFF" "#FDE725FF"
# 
# $summary_values
# [1] "2018-01-01" "2018-01-16" "2018-02-01"
# 
# $summary_colours
# [1] "#440154FF" "#21908CFF" "#FDE725FF"

colour_values(dte, n_summaries = 3, format = F)
# $colours
#  [1] "#440154FF" "#470D60FF" "#48196BFF" "#482474FF" "#472E7CFF"
#  [6] "#453882FF" "#414286FF" "#3E4B8AFF" "#3A548CFF" "#365D8DFF"
# [11] "#32658EFF" "#2E6D8EFF" "#2B758EFF" "#287D8EFF" "#25858EFF"
# [16] "#228C8DFF" "#20948CFF" "#1E9C89FF" "#20A386FF" "#25AB82FF"
# [21] "#2DB27DFF" "#39BA76FF" "#48C16EFF" "#58C765FF" "#6ACD5BFF"
# [26] "#7ED34FFF" "#92D742FF" "#A8DB34FF" "#BEDF26FF" "#D4E21BFF"
# [31] "#E9E41AFF" "#FDE725FF"
# 
# $summary_values
# [1] 17532.0 17547.5 17563.0
# 
# $summary_colours
# [1] "#440154FF" "#21908CFF" "#FDE725FF"

For categorical values use summary = TRUE to return a uniqe set of the values, and their associated colours

colour_values(sample(letters, size = 50, replace = T), summary = T)
# $colours
#  [1] "#482575FF" "#440154FF" "#345F8DFF" "#DDE318FF" "#22A884FF"
#  [6] "#43BF71FF" "#7AD151FF" "#25848EFF" "#21908CFF" "#440154FF"
# [11] "#DDE318FF" "#482575FF" "#463480FF" "#7AD151FF" "#414487FF"
# [16] "#5DC963FF" "#2A788EFF" "#DDE318FF" "#25848EFF" "#FDE725FF"
# [21] "#BCDF27FF" "#414487FF" "#FDE725FF" "#1E9C89FF" "#482575FF"
# [26] "#414487FF" "#21908CFF" "#7AD151FF" "#481466FF" "#2A788EFF"
# [31] "#481466FF" "#414487FF" "#345F8DFF" "#3B528BFF" "#481466FF"
# [36] "#5DC963FF" "#481466FF" "#5DC963FF" "#43BF71FF" "#9AD93DFF"
# [41] "#BCDF27FF" "#414487FF" "#43BF71FF" "#43BF71FF" "#FDE725FF"
# [46] "#463480FF" "#440154FF" "#2F6C8EFF" "#2A788EFF" "#2FB47CFF"
# 
# $summary_values
#  [1] "a" "c" "d" "e" "f" "g" "h" "j" "k" "m" "n" "p" "q" "r" "s" "t" "v"
# [18] "w" "x" "y" "z"
# 
# $summary_colours
#  [1] "#440154FF" "#481466FF" "#482575FF" "#463480FF" "#414487FF"
#  [6] "#3B528BFF" "#345F8DFF" "#2F6C8EFF" "#2A788EFF" "#25848EFF"
# [11] "#21908CFF" "#1E9C89FF" "#22A884FF" "#2FB47CFF" "#43BF71FF"
# [16] "#5DC963FF" "#7AD151FF" "#9AD93DFF" "#BCDF27FF" "#DDE318FF"
# [21] "#FDE725FF"

I see you support lists, but how does it work?

Basically, it’s the same as un-listing the list to create a vector of all the values, then colouring them.

So if your list contains different types, it will coerce all values to the same type and colour them.

But it returns a list of the same structure.

For example,


l <- list( x = 1:5, y = list(z = letters[1:5] ) )
colour_values( l )
# [[1]]
# [1] "#440154FF" "#482878FF" "#3E4A89FF" "#31688EFF" "#26838EFF"
# 
# [[2]]
# [[2]][[1]]
# [1] "#1F9D89FF" "#35B779FF" "#6CCE59FF" "#B4DD2CFF" "#FDE725FF"

x <- c( 1:5, letters[1:5] )
colour_values( x )
#  [1] "#440154FF" "#482878FF" "#3E4A89FF" "#31688EFF" "#26838EFF"
#  [6] "#1F9D89FF" "#35B779FF" "#6CCE59FF" "#B4DD2CFF" "#FDE725FF"

What it doesn’t do is treat each list element independently. For this you would use

lapply( l, colour_values ) 
# $x
# [1] "#440154FF" "#3B528BFF" "#21908CFF" "#5DC963FF" "#FDE725FF"
# 
# $y
# $y[[1]]
# [1] "#440154FF" "#3B528BFF" "#21908CFF" "#5DC963FF" "#FDE725FF"

What’s the performance like?

10 million numeric values

library(microbenchmark)
library(ggplot2)
library(scales)
library(viridisLite)

n <- 1e7
df <- data.frame(x = rnorm(n = n))

m <- microbenchmark(
  colourvalues = { colourvalues::colour_values(x = df$x) },
  scales = { col_numeric(palette = rgb(subset(viridis.map, opt=="D")[, 1:3]), domain = range(df$x))(df$x) },
  times = 25
)
m
# Unit: seconds
#          expr      min       lq     mean   median       uq       max neval
#  colourvalues 3.339203 3.664397 4.715676 5.026010 5.257869  6.226731    25
#        scales 6.153397 7.201396 8.834126 9.468157 9.907154 12.748400    25

autoplot(m)
# Coordinate system already present. Adding new coordinate system, which will replace the existing one.

1 million characters (26 unique values)

library(microbenchmark)
library(ggplot2)
library(scales)
library(viridisLite)

n <- 1e6
x <- sample(x = letters, size = n, replace = TRUE)
df <- data.frame(x = x)

m <- microbenchmark(
  colourvalues = { x <- colourvalues::colour_values(x = df$x) },
  scales = { y <- col_factor(palette = rgb(subset(viridis.map, opt=="D")[, 1:3]), domain = unique(df$x))(df$x) },
  times = 25
)
m
# Unit: milliseconds
#          expr      min       lq     mean   median       uq       max neval
#  colourvalues 447.7285 488.8479 519.8084 507.0384 526.8926  686.7764    25
#        scales 888.0087 936.6404 972.2149 960.9733 997.1543 1133.9043    25

autoplot(m)
# Coordinate system already present. Adding new coordinate system, which will replace the existing one.

colourvalues

What does it do?

Why did you build it?

Why do you spell colour with a ‘u’?

How do I install it?

How can I make use of it in my package?

Do you have any examples?

256 numbers mapped to a colour

5000 numbers on a non-linear scale

1000 random numbers

Are there only viridis palettes?

Do I have to use the in-built palettes?

Do you support ‘alpha’ values

Some of my plotting functions don’t support alpha, can I exclude it?

Can I get a summary of colours to use in a legend?

I see you support lists, but how does it work?

What’s the performance like?

Copy Link

Version

Install

Monthly Downloads

Version

License

Maintainer

Last Published

Functions in colourvalues (0.3.1)