Combine Categorical and Gradient Fill in Geospatial - R

问题

I'm trying to fill a combined categorical and continuous variable on a map. So, for instance, in my minimally reproducible example below, say I want to display the number of KrispyKreme Donut shops in each county, which is generally a continuous variable I want to fill on a gradient. But I also have counties that forbid KrispyKremes indicated by a "-1" and those that have them under construction "-2". I want to display these in a different color not mapped on the gradient. I also have NA in my real data.

--What I have so far:

library(sf)
library(ggplot2)

nc <- st_read(system.file("shape/nc.shp", package="sf"))
nc$Status<-rep(c(-2,-1,runif(8)), 10)

ggplot(nc) + 
  geom_sf(aes(fill=Status),color = "black") + 
  coord_sf(datum = NA) + 
  theme_minimal()

It breaks if I add the following line, obviously. So, I know I have the syntax wrong but it indicates what I want to do in as best as I can figure code for this

  scale_fill_manual(breaks= c("-2","-1", >=0),values = c("blue", "yellow", scale_fill_viridis()))

Any help is much appreciated, I've been on this all day.

回答1:

You will need to cut your continuous variable into different categories.

library(sf)
library(ggplot2)
library(dplyr)

# Set seed for reproducibility
set.seed(122)

nc <- st_read(system.file("shape/nc.shp", package="sf"))
nc$Status<-rep(c(-2,-1,runif(8)), 10)

First, check the distribution of your variable.

nc %>%
  filter(Status >= 0) %>%
  pull("Status") %>%
  summary()
#     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
# 0.002789 0.153144 0.602395 0.491287 0.735787 0.906851

I decided to cut the variable based on the quantile as follows.

nc2 <- nc %>%
  mutate(Status2 = case_when(
    Status == -2 ~ "-2",
    Status == -1 ~ "-1",
    Status >= 0 & Status < 0.15 ~ "0 - 0.15",
    Status >= 0.15 & Status < 0.6 ~ "0.15 - 0.6",
    Status >= 0.6 & Status < 0.75 ~ "0.6 - 0.75",
    Status >= 0.75                ~ "0.75 - 0.91"
  ))

Now Status2 is a categorical variable. We can plot it and use scale_fill_manual to provide colors. Notice that we need to provide the color code in the values argument. viridis::viridis(4) is to generate four colors based on the viridis.

ggplot(nc2) + 
  geom_sf(aes(fill=Status2),color = "black") + 
  coord_sf(datum = NA) + 
  theme_minimal() +
  scale_fill_manual(values = c("blue", "yellow", viridis::viridis(4)))

回答2:

Many thanks for this. The way Status2 was constructed above makes it a character variable. I would like to plot a categorical variable instead. The code below generates a factor variable instead (Status3) and plots it in the map. It works.

library(sf) 
library(ggplot2) 
library(dplyr)

nc <- st_read(system.file("shape/nc.shp", package="sf")) 
nc$Status<-rep(c(-2,-1,runif(8)), 10)

nc3 <- nc %>%
  mutate(Status3 = factor(ifelse(Status>0,1,0)))

ggplot(nc3) + 
  geom_sf(aes(fill=Status3),color = "black") + 
  coord_sf(datum = NA) + 
  theme_minimal()

However, when I try to do apply the same rationale (construct a factor variable based on a continuous variable and plot it a map) to my code I get an error.

Error in if (type == "point") { : argument is of length zero

My code is below. The code works when plotting the continuous variable but not when plotting the factor variable. Does anyone know why?

# plotting continuous variable: WORKS FINE
ggplot(CS_mun_shp)+
  geom_sf(aes(geometry=geometry,
              fill=ppc_sih),
          color=NA) 

# constructing factor variable
CS_mun_shp2 <- CS_mun_shp %>%
  mutate(cs_above40=factor(ifelse(ppc_sih>=0.4,1,0), 
                           levels=c(0:1), 
                           labels=c('below 40%','above 40%')))

# plotting factor variable: GENERATES ERROR  
ggplot(CS_mun_shp2)+
  geom_sf(aes(geometry=geometry,
              fill=cs_above40),
          color=NA)

The only difference between my code and the reproduceable example above is that I need to specify geometry inside aes(), otherwise I get another error.

来源：https://stackoverflow.com/questions/53126643/combine-categorical-and-gradient-fill-in-geospatial-r

标签

ggplot2

geospatial