GGplot Theme Sampler: Choosing a Theme

So Many Themes! Which One to Use?

A small selection of themes available for ggplot2

A Sample of Themes to Help in Selecting the Right One

Sooner or later, the times comes to move beyond the default ggplot2 grey theme and explore other options. But which theme to use? And When?

Here is a collection of ggplot charts with different themes applied, intended to provide some helpful comparisons and make it easier to choose the best option for the situation.

The focus is on themes that come with the ggplot package, but also included are some examples from ‘ggthemes’ and ‘ggpubr’ packages. Consideration could be extended to other packages like ‘hrbrthemes’, ‘ggthemr’, ‘themetron’ or others. Although the possibilities are endless, realistically, many of the themes are variations on existing themes/concepts and it may be just as easy to work with a built-in theme and tweak it to your own desired flavour. Some examples and thoughts on this at the end.

Theme sources used:

• ggplot2 built-in: if none of the below specified.
• ggthemes library: identified by ‘ggt’ in chart titles below.
• ggpubr library: identified by ‘ggp’ in chart titles below.
  • the ‘transparent’ theme suppresses title by default, so extra modification was needed. if used ‘out of the box’, will not have a title, even if title is specified.

Colors:

Default ggplot colors used for expediency - exploring color palettes is another adventure altogether.

First, the data

Using global_economy data set from tsibbledata package because it is characteristics of data I typically work with:

• time series
• additional dimension (category, product, traffic source, region…)
• multiple measures on different scales, for variety

Of course, this is not exhaustive exploration of the variations, but intended as an illustrative cross-section.

# data set with time dimension, at least one other dimension, at least one metric
chart_data <- tsibbledata::global_economy
# add metric for GDP per capita
chart_data <- chart_data %>% mutate(
  GDP_per_cap = GDP / Population
)
# isolate country for use case
chart_data_single <- chart_data %>% filter(Code=='CAN')
# filter for select countries to make manageable for use cases
chart_data_select <- chart_data %>% filter(Code %in% c("CAN","USA","BRA","JPN","IND","NGA")) 

Function to take chart info and process themes

A reusable function for ease of sampling different themes with different chart types.

# function to take chart info as input, apply themes, return variations with themes
chartThemeSampler <- function(chart_example) {
  c_dflt <- chart_example + labs(title = "Default (aka Gray/Grey)")
  c_m <- chart_example + labs(title = "Minimal") + theme_minimal()
  c_l <- chart_example + labs(title = "Light") + theme_light()
  c_drk <- chart_example + labs(title = "Dark") + theme_dark()
  c_c <- chart_example + labs(title = "Classic") + theme_classic()
  c_bw <- chart_example + labs(title = "BW") + theme_bw()
  # ggthemes
  c_econw <- chart_example + labs(title = "economist_white (ggt)")+
            theme_economist_white()
  c_few <- chart_example + labs(title = "Few (ggt)") + theme_few()
  c_five <- chart_example + labs(title = "FiveThirtyEight (ggt)") + theme_fivethirtyeight()
  c_gdoc <- chart_example + labs(title = "GDocs (ggt)") + theme_gdocs()
  c_hc <- chart_example + labs(title = "HC (ggt)") + theme_hc()
  c_sol2 <- chart_example + labs(title = "Solarized  2(ggt)") + theme_solarized_2()
  c_tufte <- chart_example + labs(title = "theme_tufte (ggthemes)") + theme_tufte()
  c_wsj <- chart_example + labs(title = "WSJ (ggt)") + theme_wsj()
  ## ggpubr
  c_clev <- chart_example + labs(title = "Cleveland (ggp)") + theme_cleveland()
    # theme_transparent requires extra code to display title
  c_transp <- chart_example + labs(title = "Transparent (ggp)") + theme_transparent() + 
    theme(plot.title = element_text())
  
  # combine
  cts <- list(c_dflt, c_m, c_l, c_drk, c_c, c_bw, 
              c_econw, c_few, c_five, c_gdoc, c_hc, c_sol2, c_tufte, c_wsj,
              c_clev, c_transp)
  for(i in 1:length(cts)) {
    print(cts[[i]])
  }
}

Histograms

# set up chart for applying themes
chart_example <- chart_data_select %>% ggplot(aes(x=Growth))+geom_histogram()
# use function to apply themes to the charts and display
chartThemeSampler(chart_example)

Line Charts

chart_example <- chart_data_single %>% ggplot(aes(x=Year, y=Growth))+geom_line()
chartThemeSampler(chart_example)

Line charts w/Legend

chart_example <- chart_data_select %>% ggplot(aes(x=Year, y=Growth, col=Code))+geom_line()
chartThemeSampler(chart_example)

Column (Bar) Charts

chart_example <- chart_data_single %>% ggplot(aes(x=Year, y=Growth))+geom_col()
chartThemeSampler(chart_example)

Column Charts - Facets

chart_example <- chart_data_select %>% 
  filter(Year >= 2000 & Code %in% c("BRA","CAN","IND")) %>%
  ggplot(aes(x=Year, y=Growth))+geom_col()+
  facet_grid(Code~.)
chartThemeSampler(chart_example)

Boxplots

chart_example <- chart_data_select %>% 
  ggplot(aes(x=Code, y=Growth))+geom_boxplot()
chartThemeSampler(chart_example)

Scatterplots

chart_example <- chart_data_single %>% ggplot(aes(x=GDP_per_cap, y=Growth))+geom_point()
chartThemeSampler(chart_example)

Scatterplots - Facets

chart_example <- chart_data_select %>% filter(Code %in% c("BRA","CAN","IND")) %>%
  ggplot(aes(x=Exports, y=Imports))+geom_point()+
  facet_grid(Code~.)
chartThemeSampler(chart_example)

Mods / Experiments

Clearly, there are lots of options readily available to suit most situations. All of these themes have their strengths and limitations.

Still, it is entirely possible that none of the ‘out-of-the-box’ themes may fit exactly with what you want. Here are some examples of minor modifications to built-in themes to see how they play out across different chart types.

Chart Setup & Function for mods

Chart setup

A collection of charts to apply themes to, as opposed to applying various themes to a specific chart, as above.

chart_line <- chart_data_single %>% ggplot(aes(x=Year, y=GDP)) + geom_line()
chart_col <- chart_data_single %>% ggplot(aes(x=Year, y=GDP)) + geom_col()
chart_scat <- chart_data_select %>% ggplot(aes(x=Exports, y=Growth)) + geom_point()
chart_box <- chart_data_select %>% ggplot(aes(x=Code, y=Growth)) + geom_boxplot()
# facets
chart_col_facet <- chart_data_select %>% 
  filter(Year >= 2000 & Code %in% c("JPN","NGA","USA")) %>%
  ggplot(aes(x=Year, y=Growth))+
  geom_col()+
  facet_grid(Code~.)
chart_scat_facet <- chart_data_select %>% filter(Code %in% c("BRA","CAN","IND")) %>%
  ggplot(aes(x=Imports, y=Growth)) + geom_point() +
  facet_grid(Code~.)

Function for processing themes

# takes list of chart settings plus additional theme definition as inputs and combines
chartMod <- function(charts, chart_mods, theme_def) {
  for(c in 1:length(charts)) {
    chart_modo <- charts[[c]] + chart_mods + theme_def
    print(chart_modo)
  }
}

Light without grid lines

• theme_light, with border
• no gridlines
• geom_hline for line at 0 on y-axis

# set list of charts
charts <- list(chart_line, chart_col, chart_scat, chart_box, chart_col_facet, chart_scat_facet)
# add chart modifications
chart_mods <- geom_hline(yintercept = 0, color='grey90')
# set theme to apply
theme_def <- theme_light() + theme(panel.grid = element_blank())
# run function to show charts
chartMod(charts, chart_mods, theme_def)

Light wth major gridlines only

Major gridlines for reference with color lightened and size reduced. 0 line for extra context, distinguished from gridlines.

charts <- list(chart_line, chart_col, chart_scat, 
               chart_box, chart_col_facet, chart_scat_facet)
chart_mods <- geom_hline(yintercept = 0, color = 'grey60') 
theme_def <- theme_light() + theme(panel.grid.major = 
                                     element_line(color = 'grey80', linewidth = 0.1),
                                  panel.grid.minor = element_blank())
chartMod(charts, chart_mods, theme_def)

Few with reference line

• Reference line at 0 for context.
• Essentially same as ‘Light’ without gridlines, above, but has some cleaner labelling, notable on facets.

charts <- list(chart_line, chart_col, chart_scat, 
               chart_box, chart_col_facet, chart_scat_facet)
chart_mods <- geom_hline(yintercept = 0, color = 'grey70') 
theme_def <- theme_few()
chartMod(charts, chart_mods, theme_def)

Transparent with border

• Extremely minimal, could be useful in showing general pattern, without details.
• Border may help to provide some sense of ‘boundary area’

charts <- list(chart_line, chart_col, chart_scat, 
               chart_box, chart_col_facet, chart_scat_facet)
chart_mods <- geom_hline(yintercept = 0, color = "grey90") 
theme_def <- theme_transparent()+theme(panel.border = element_rect(size=1, color='grey90', fill=NA))
chartMod(charts, chart_mods, theme_def)

Solarized_2 with major grid only

‘0’ line highlighted.

charts <- list(chart_line, chart_col, chart_scat, 
               chart_box, chart_col_facet, chart_scat_facet)
chart_mods <- geom_hline(yintercept = 0, color="white") 
theme_def <- theme_solarized_2() + theme(panel.grid.minor = element_blank())
chartMod(charts, chart_mods, theme_def)

Closing suggestions

• consider your objectives, audience and priorities:
  • if time is of the essence, a built-in chart or even default may be good enough
  • if design is important to your audience, you may want to go the extra distance to craft something more specialized
  • if looking to create a personal or professional brand, may want to create your own template that you can use repeatedly over time and different situations
• keep things as simple as possible, but no simpler: provide enough guidance to direct and assist the eye
• consider reference lines in addition to / instead of gridlines to provide easier anchoring

Above all else, for the sake of your data consumers…keep on visualizing!