20

I have a ggplot with one facet for each US state. I would like to arrange these facets in the shape of the US with a ragged border (ordered like the second map, but without Hawaii or Alaska).

To do this, I created a state-level factor variable that is ordered by US state, as read from left to right on a map. This factor also contains "space holders" for blank facets that I'd like to remove. I followed the advice of this post (see the Edit to the provided answer) but names(g$grobs) is NULL, so I can't implement their answer. Any ideas what I can do?

Here is my code:

library(ggplot2)
library(fivethirtyeight)
library(dplyr)
library(gridExtra)

data("police_deaths")
police_deaths_count <- police_deaths %>% arrange(state, -year) %>% group_by(state, year) %>% count()
police_deaths_count <- police_deaths_count %>% arrange(state, -year) %>% 
  filter(year %in% c(1970:2015) & !state %in% c("AK", "HI", "US", "GU", "MP", "PR", "RR", "TR", "VI"))

police_deaths_count$state.name <- state.name[match(police_deaths_count$state, state.abb)]
police_deaths_count$state.name[police_deaths_count$state == "DC"] <- "Washington DC"

police_deaths_count$state.reorder <- factor(police_deaths_count$state.name,
                                         levels = c("e1", "e2", "e3", "e4", "e5", "e6", "e7", "e8", "e9", "e10", "Maine",
                                                    "e11", "e12", "e13", "e14", "e15", "e16", "e17", "e18", "e19", "Vermont", "New Hampshire",
                                                    "Washington", "Idaho", "Montana", "North Dakota", "Minnesota", "Illinois", "Wisconsin", "Michigan", "New York", "Massachusetts", "Rhode Island",
                                                    "Oregon", "Nevada", "Wyoming", "South Dakota", "Iowa", "Indiana", "Ohio", "Pennsylvania", "New Jersey", "Connecticut", "e20",
                                                    "California", "Utah", "Colorado", "Nebraska", "Missouri", "Kentucky", "West Virginia", "Virginia", "Maryland", "Washington DC", "e21",
                                                    "e22", "Arizona", "New Mexico", "Kansas", "Arkansas", "Tennessee", "North Carolina", "South Carolina", "Delaware", "e23", "e24",
                                                    "e25", "e26", "e27", "Oklahoma", "Louisiana", "Mississippi", "Alabama", "Georgia", "e28", "e29",
                                                    "e30", "e31", "e32", "e33", "Texas", "e34", "e35", "e36", "e37", "Florida"))

police_deaths_count2 <- police_deaths_count %>% filter(!(state=="NY" & year==2001))

plot1 <- ggplot(subset(police_deaths_count2, is.na(state.name)==F), #take away 9-11 peak to see trends without it
               aes(y = n, x = year)) +
  geom_line() + 
  facet_wrap( ~ state.reorder, ncol = 11, drop = F) +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  ylab("Count of police deaths") +
  xlab("Year")

#the order of these facets is what I want. From here, I'd like to display the facets e1, e2, ..., e37 as completely blank by removing their facet strips and panels.
plot1

#A SO post (next line) provides a potential solution, but it doesn't work for me
#https://stackoverflow.com/questions/30372368/adding-empty-graphs-to-facet-wrap-in-ggplot2 
g <- ggplotGrob(plot1)
names(g$grobs) #this is NULL so I can't implement the SO answer.
g$layout$name
Community
  • 1
  • 1
corinne r
  • 290
  • 2
  • 11

3 Answers3

19

A hack-ish option would be to create unique blank strip labels for the empty facets, so that they can be used as placeholders, but without creating any visible strip labels. It would probably also be better to use state abbreviations instead of full names, but I haven't done that here. Here's an example:

library(ggplot2)
library(fivethirtyeight)
library(dplyr)
library(gridExtra)

data("police_deaths")
police_deaths_count <- police_deaths %>% arrange(state, -year) %>% group_by(state, year) %>% count()
police_deaths_count <- police_deaths_count %>% arrange(state, -year) %>% 
  filter(year %in% c(1970:2015) & !state %in% c("AK", "HI", "US", "GU", "MP", "PR", "RR", "TR", "VI"))

# Create unique blank strip labels for empty facets
bl = sapply(1:37, function(n) paste(rep(" ",n),collapse=""))

police_deaths_count$state.name <- state.name[match(police_deaths_count$state, state.abb)]
police_deaths_count$state.name[police_deaths_count$state == "DC"] <- "Washington DC"

police_deaths_count$state.reorder <- factor(police_deaths_count$state.name,
                                            levels = c(bl[1:10], "Maine",
                                                       bl[11:19], "Vermont", "New Hampshire",
                                                       "Washington", "Idaho", "Montana", "North Dakota", "Minnesota", "Illinois", "Wisconsin", "Michigan", "New York", "Massachusetts", "Rhode Island",
                                                       "Oregon", "Nevada", "Wyoming", "South Dakota", "Iowa", "Indiana", "Ohio", "Pennsylvania", "New Jersey", "Connecticut", bl[20],
                                                       "California", "Utah", "Colorado", "Nebraska", "Missouri", "Kentucky", "West Virginia", "Virginia", "Maryland", "Washington DC", bl[21],
                                                       bl[22], "Arizona", "New Mexico", "Kansas", "Arkansas", "Tennessee", "North Carolina", "South Carolina", "Delaware", bl[23:24],
                                                       bl[25:27], "Oklahoma", "Louisiana", "Mississippi", "Alabama", "Georgia", bl[28:29],
                                                       bl[30:33], "Texas", bl[34:37], "Florida"))


police_deaths_count2 <- police_deaths_count %>% filter(!(state=="NY" & year==2001))

plot1 <- ggplot(subset(police_deaths_count2, is.na(state.name)==F), #take away 9-11 peak to see trends without it
                aes(y = n, x = year)) +
  geom_line() + 
  facet_wrap( ~ state.reorder, ncol = 11, drop = F, strip.position="bottom") +
  theme_classic() +
  theme(axis.text.x = element_blank(),
        strip.background=element_blank(),
        axis.line=element_blank(),
        axis.ticks=element_blank()) +
  ylab("Count of police deaths") +
  xlab("Year (1970 - 2015)")

enter image description here

eipi10
  • 91,525
  • 24
  • 209
  • 285
8

The geofacet package should work nicely: https://github.com/hafen/geofacet

Phil
  • 7,287
  • 3
  • 36
  • 66
  • `geofacet` came out after I asked this question, but I use it now myself as well :) – corinne r Oct 19 '17 at 17:46
  • 1
    It did, but it's always useful to update these questions for people who stumble onto your question. :) – Phil Oct 19 '17 at 17:48
1

There is a function called glyphs in the GGally package which can make these.

David LeBauer
  • 31,011
  • 31
  • 115
  • 189
Dianne Cook
  • 139
  • 3