Pubtheme
`ggplot` and `plotly` themes for creating publication quality visualizations for web and mobile.
Install / Use
/learn @bmacGTPM/PubthemeREADME
pubtheme
<!-- badges: start --> <!-- badges: end -->The package pubtheme contains a ggplot theme theme_pub for
creating data journalism-style data visualizations with color palettes
and formatting similar to those used by media organizations like BBC, NY
Times, and ESPN. Several templates for scatter plot, line plots, etc.,
are provided below for easy copying/pasting. A helpful function pub
can be used to reduce then amount of code that needs to be typed when
the user is comfortable with the default settings.
Organization-specific color palettes and logos can be used as well via
the package orgthemes. See https://github.com/bmacGTPM/orgthemes.
Installation
If you don’t have the package devtools, install it using
install.packages('devtools'). If you have devtools, you can install
the GitHub version of pubtheme with:
devtools::install_github("bmacGTPM/pubtheme")
If you get an error about download method, try changing this option before installing.
options(download.file.method = 'libcurl')
Load the package using
library(pubtheme)
as usual. The theme will change some of your ggplot defaults the first time you use it. To change them back, restart R, or use
restore.ggplot.defaults()
at any time. It is recommended that you update your version of
tidyverse and especially ggplot2 to use this package.
Scatter plot
dg = mtcars %>%
select(wt, mpg, cyl) %>%
mutate(Cylinders = as.factor(cyl)) %>%
rename(MPG = mpg)
title = "Title in Upper Lower"
g = ggplot(dg,
aes(x = wt,
y = MPG,
color = Cylinders,
size = MPG)) +
geom_point() +
labs(title = title,
subtitle = 'Optional Subtitle In Upper Lower',
caption = "Optional caption with more info, X handle, or shameless promotion of pubtheme",
x = 'Horizontal Axis Label in Upper Lower',
y = 'Vertical Axis Label in Upper Lower') +
scale_x_continuous(limits = c(0, 6),
breaks = c(0, 3, 6),
oob = squish,
labels = comma) +
scale_y_continuous(limits = c(0,40),
breaks = c(0,20,40),
oob = squish,
labels = comma) +
scale_size(range = c(2, 6)) +
coord_cartesian(clip = 'off',
expand = F) +
theme_pub(type = 'scatter')
print(g)
## Save to a file using base_size = 36
gg = g +
scale_size(range = c(6, 18)) +
theme_pub(type = 'scatter',
base_size = 36)
ggsave(filename = paste0("img/", gsub("%", " Perc", title), ".jpg"),
plot = gg,
width = 20, ## do not change
height = 20, ## can change from 20 if desired
units = 'in', ## do not change
dpi = 72) ## do not change
g.scatter = g ## save for later
You must have a subfolder called img in order for the ggsave chunk
above to work.
Note that the default is base_size = 12, which works well when viewing
the image in RStudio. Use ggsave and base_size = 36 when saving an
image instead of exporting from the RStudio viewer. Do not adjust the
width = 20 in ggsave.
Do not change width = 20, units = 'in', or dpi = 72. Height
can be adjusted if desired. A square image is often preferred, so when
in doubt, keep height at 20.
Upper Lower means First letter of each word is capitalized. The option
expand = FALSE removes the default padding. The option
breaks=c(0, 3, 6) sets 3 lines at left, middle, and right. You can
certainly add lines if there is a reason to, but when in doubt you can
stick with just 3 lines (left/middle/right) only. Similarly, for the
y-axis, top/middle/bottom only.
You’ll notice a scale_size(range = c(6, 18) when preparing the plot to
be saved. If you are using size inside aes(), you’ll need that
change the scale, otherwise the points will be too small.
pub function
We explicitly specified the scale* and coord_cartesian above. To use
the default settings in pubtheme, you can save a lot of typing by
using the function pub, which applies theme_pub and also
automatically adds scale* and coord* similar to above. The function
also aligns the legend with the left side of the title and subtitle.
title = "Title in Upper Lower"
g = ggplot(dg,
aes(x = wt,
y = MPG,
color = Cylinders,
size = MPG)) +
geom_point() +
labs(title = title,
subtitle = 'Optional Subtitle In Upper Lower',
caption = "Optional caption giving more info, X handle, or shameless promotion of pubtheme",
x = 'Horizontal Axis Label in Upper Lower',
y = 'Vertical Axis Label in Upper Lower')
g %>%
pub(xlim = c(0, 6),
ylim = c(0, 40))
## Save to a file using base_size = 36
gg = g %>%
pub(xlim = c(0, 6),
ylim = c(0, 40),
base_size = 36)
ggsave(filename = paste0("img/",
gsub("%", " Perc", title),
".jpg"),
plot = gg,
width = 20, ## do not change
height = 20, ## can change from 20 if desired
units = 'in', ## do not change
dpi = 72) ## do not change
To save to a file, we again need base_size = 36. We simply copy and
paste the pub code and add base_size = 36.
Correlation plot
Let’s make a correlation plot in the look of pubtheme. Since
corrplot gives a corrplot object, pub can’t be used with it, so we
have to make a correlation plot from scratch using geom_tile.
## choose order of variables
cols = sort(unique(colnames(mtcars)))
corr <- round(cor(mtcars[,cols]), 2)
head(corr,2)
#> am carb cyl disp drat gear hp mpg qsec vs wt
#> am 1.00 0.06 -0.52 -0.59 0.71 0.79 -0.24 0.60 -0.23 0.17 -0.69
#> carb 0.06 1.00 0.53 0.39 -0.09 0.27 0.75 -0.55 -0.66 -0.57 0.43
dg = corr %>%
as.data.frame() %>%
## longer format
rownames_to_column(var = 'x') %>%
pivot_longer(cols = -x,
names_to = 'y',
values_to = 'value') %>%
## order them however you'd like
mutate(x = factor(x, levels = rev(sort(unique(x)))),
y = factor(y, levels = sort(unique(y))))
title = "Title in Upper Lower"
g = ggplot(dg) +
geom_tile(aes(x = x,
y = y,
fill = value),
linewidth = 0.4,
show.legend = T,
color = pubdarkgray) +
scale_fill_gradientn(colors = c('red4',
pubred,
publightred,
'white', ## or 'pubbackgray'
publightblue,
pubblue,
'navy'),
na.value = pubmediumgray, ## same color as below
oob = squish,
breaks = c(-1, 0, 1),
limits = c(-1, 1)) +
labs(title = title,
subtitle = 'Optional Subtitle In Upper Lower',
caption = "Optional caption giving more info, X handle, or shameless promotion of pubtheme",
x = 'Day (Optional Axis Label in Upper Lower)',
y = NULL, ## Optional
fill = 'Value')
g %>%
pub(type = 'grid') +
theme(axis.text.x.top = element_text(angle = 90,
hjust = 0,
vjust = 0.3))
gg = g %>%
pub(type = 'grid',
base_size = 36)
ggsave(filename = paste0("img/", gsub("%", " Perc", title), ".jpg"),
plot = gg,
width = 20, ## do not change
height = 24, ## can change from 20 if desired. We use 12 here to make the tiles square
units = 'in', ## do not change
dpi = 72) ## do not change
Adding rectangles around clusters, as is often done in corrplot.
# ## note that the function corrRect.hclust is used there
# corrplot::corrplot
#
# ## note that this converts corr to a distance matrix, and then uses cutree to get the clusters
# corrplot::corrRect.hclust
corr = cor(mtcars)
ds = as.dist(1-corr)
dh = hclust(ds)
hc = cutree(dh, k = 4)
dr = data.frame(order = dh$order,
cluster = hc) %>%
mutate(cluster = factor(cluster,
levels = c(1,3,4,2))) %>% ## this order looks better
arrange(cluster, order) %>%
mutate(x = 1:n(),
y = n():1) ## gotta reverse the order
dg = corr %>%
as.data.frame() %>%
## longer format
rownames_to_column(var = 'x') %>%
pivot_longer(cols = -x,
names_to = 'y',
values_to = 'value') %>%
## order them the same way as dr
mutate(x = factor(x, levels = rownames(dr)),
y = factor(y, levels = rev(rownames(dr))))
polys = dr %>%
group_by(cluster) %>%
summarise(xmin = min(x),
xmax = max(x),
ymin = min(y),
ymax = max(y)) %>%
mutate(xmin = xmin - 0.5,
xmax = xmax + 0.5,
ymin = ymin - 0.5,
ymax = ymax + 0.5,
linewidth = 1.5)
title = "Title in Upper Lower"
g = ggplot(dg) +
## Tiles
geom_tile(aes(x = x,
y = y,
fill = value),
linewidth = 0.4,
show.legend = T,
color = pubdarkgray) +
## Rectangles
geom_rect(data = polys,
aes(xmin = xmin,
xmax = xmax,
ymin = ymin,
ymax = ymax,
gro
