CODEWITHSUNDEEP
rggplot2
user51966
user51966

Reputation: 1057

ggplot2: Change the order of `color`

I have a figure like below and I would like to change the order of colored lines from blue->green->red to r->g->b, so that it is the same as the legend's order. I have found several tutorials to change the order of legends, but I want to keep its order in this case (since they are 1, 2, and 3).

Figure

Here is the code to generate the data and figure.

population_num <- 100
population <- tibble(
    gender = as.numeric(rbinom(population_num, 1, 0.5)),
    age=rnorm(population_num, mean=50, sd=20),
    score=rnorm(population_num, mean=80, sd=30),
    setid=sample(c(1,2,3), size=population_num, replace=T)
    )
temp <- population %>%
  group_by(setid) %>%
  do(model1 = tidy(lm(score ~ age, data = .)),
     model2 = tidy(lm(score ~ age + gender, data = .))) %>% 
  gather(model_name, model, -setid) %>%                     
  unnest() %>%                                              
  filter(term == "age")                                     

interval1 <- -qnorm((1-0.9)/2)

ggplot(temp, aes(colour = as.factor(setid))) +
  geom_hline(yintercept = 0, colour = gray(1/2), lty = 2) +
  geom_linerange(aes(x = model_name, ymin = estimate - std.error*interval1,
                     ymax = estimate + std.error*interval1),
                 lwd = 1, position = position_dodge(width = 1/2)) +
 scale_x_discrete(limits=c("model2", "model1"), labels=c("M2", "M1")) +
  coord_flip()

(This question is once asked in Japanese Stackoverflow, but couldn't get answers.)

Upvotes: 2

Views: 2763

Answers (1)

alistaire
alistaire

Reputation: 43334

You can change the width parameter of position_dodge to a negative. This does produce a warning:

Warning message:
position_dodge requires non-overlapping x intervals

but plots fine:

ggplot(temp, aes(colour = as.factor(setid))) +
    geom_hline(yintercept = 0, colour = gray(1/2), lty = 2) +
    geom_linerange(aes(x = model_name, ymin = estimate - std.error*interval1,
                       ymax = estimate + std.error*interval1),
                   lwd = 1, position = position_dodge(width = -1/2)) +
    scale_x_discrete(limits=c("model2", "model1"), labels=c("M2", "M1")) +
    coord_flip()

Upvotes: 3

Related Questions