Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/njtierney/palap
symmetric (reflective, palindromic) colour palettes (rev(palap) == palap)
https://github.com/njtierney/palap
Last synced: 2 months ago
JSON representation
symmetric (reflective, palindromic) colour palettes (rev(palap) == palap)
- Host: GitHub
- URL: https://github.com/njtierney/palap
- Owner: njtierney
- License: other
- Created: 2019-09-13T03:24:23.000Z (about 5 years ago)
- Default Branch: master
- Last Pushed: 2020-10-02T09:22:08.000Z (about 4 years ago)
- Last Synced: 2024-08-13T07:15:33.925Z (4 months ago)
- Language: R
- Homepage:
- Size: 680 KB
- Stars: 17
- Watchers: 3
- Forks: 0
- Open Issues: 4
-
Metadata Files:
- Readme: README.Rmd
- Changelog: NEWS.md
- License: LICENSE
Awesome Lists containing this project
- jimsghstars - njtierney/palap - symmetric (reflective, palindromic) colour palettes (rev(palap) == palap) (R)
README
---
output: github_document
---
```{r setup, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
# palap
[](https://www.tidyverse.org/lifecycle/#experimental)
[](https://ci.appveyor.com/project/njtierney/palap)
[](https://travis-ci.org/njtierney/palap)
[](https://codecov.io/gh/njtierney/palap?branch=master)
The goal of palap is to provide "symmetric" colour palettes. Broadly speaking, there are three categories of colour palette:
1\. **Qualitative**: categorical variables
2\. **Sequential**: low to high numeric values
3\. **Diverging**: negative to positive values
However, a fourth colour palette is needed, to cover cases where things diverge, but maintain some constant value/meaning as they get further from the center point. The proposed name for this is a "symmetric" colour palette:
4\. **Symmetric**: Diverging values with equivalent meaning as they get further from the diverging point.
At the moment this package exists as a proof of concept, and a place to explore this idea. It is currently drawing heavily from [scico](https://github.com/thomasp85/scico).
## Installation
You can install the released version of `palap` from github with:
``` r
# install.packages("remotes")
remotes::install_github("njtierney/palap")
```
# Example
We can see a normal sequential colourplot such as the Yellow Green Blue palette here:
```{r example}
library(palap)
library(prismatic)
library(paletteer)
regular <- paletteer_d(package = "RColorBrewer", palette = "YlGnBu", n = 7)
# use the colour and plot functions from prismatic
plot(colour(regular))
```
Using `palap`, we can create a symmetric version of this colour palette
```{r example-palap}
# now show this with palap
sym_pal <- palap(7)
plot(colour(sym_pal))
```
We can also change the direction of the colourscale
```{r example-reverse}
plot(colour(palap(7, direction = -1)))
```
Note that when an even number is given, the colour palette will be "doubled" - like so:
```{r palap-even}
plot(colour(palap(8)))
plot(colour(palap(8, direction = -1)))
```
# Using other colour palettes
Thanks to [`paletteer`](https://github.com/EmilHvitfeldt/paletteer) by [Emil Hvitfeldt](https://github.com/EmilHvitfeldt), you can use a wide variety of colour palettes from other packages.
By default it uses `RColorBrewer` and the Yellow Green Blue colour (YlGnBu) colour palette.
You could change this so that you can consider the Blue Purple (BuPu) palette:
```{r show-blue-purple}
plot(colour(palap(7, palette = "BuPu")))
```
For symmetric colour scales we recommend that you use colour scales that are sequential.
# Future work
- Generalise across other packages, only using sequential colour palettes.
- Generate your own symmetric colour palette from a given vector of colours - perhaps similarly to [paletti](https://github.com/EdwinTh/paletti).
# example with lm
You might want to fit a linear model and add two levels of confidence intervals.
```{r long-example-of-lm}
lm_fit <- lm(speed ~ dist, cars)
conf_int <- function(x, conf_level){
predict(x,
interval = c("confidence"),
level = conf_level) %>%
tibble::as_tibble() %>%
setNames(c("fit",
paste0("lwr_",conf_level*100),
paste0("upr_",conf_level*100)))
}
library(dplyr)
library(tidyr)
library(tibble)
lm_conf <- conf_int(lm_fit, 0.95) %>%
bind_cols(conf_int(lm_fit, 0.80)) %>%
bind_cols(cars) %>%
select(-fit1) %>%
pivot_longer(cols = c(lwr_95:upr_80),
names_to = "conf_var",
values_to = "conf_val") %>%
mutate(conf_var = factor(x = conf_var,
levels = c("lwr_95",
"lwr_80",
"upr_80",
"upr_95"),
ordered = TRUE))
```
So your data ends up looking like this:
```{r show-lm-conf}
lm_conf
```
But then when you plot it as some separate lines for each level of confidence, you get this:
```{r gg-lm-no-palap}
library(ggplot2)
gg_conf <-
ggplot(lm_conf,
aes(x = dist,
y = speed)) +
geom_point() +
geom_line(aes(y = fit)) +
geom_line(aes(y = conf_val,
colour = conf_var))
# without palap - the bands aren't really ordered in a meaningful way
gg_conf
```
But using palap, you get a nicer colourscale
```{r gg-palap}
# with palap - scales are ordered in a way that makes sense
gg_conf +
scale_colour_palap_d()
```
# Example from `brolgar`
```{r brolgar}
library(dplyr)
library(brolgar)
heights_near <- key_slope(heights,
height_cm ~ year) %>%
keys_near(key = country,
var = .slope_year) %>%
left_join(heights, by = "country")
library(ggplot2)
p <- ggplot(heights_near,
aes(x = year,
y = height_cm,
group = country,
colour = stat)) +
geom_line()
p
p + scale_colour_palap_d()
p + scale_colour_palap_d(direction = -1)
```
# Acknowledgements
Thank you to [Di Cook](http://dicook.org/) for pointing out the need for this kind of colour palette, it is one of those things that once you notice, you cannot unsee, and that's so cool. Thanks also to [Alison Hill](https://alison.rbind.io/) for pointing out that I could call this "symmetric" instead of "reflective" - to me this is much clearer.