1

I have the following data frame resulting from simulations of ODEs with different parameter sets, e.g.

df <- data.frame(t = rep(seq(0,4), 4),
                 x1 = c(1.2*seq(1,5), 1.3*seq(1,5), 1.4*seq(1,5), 1.5*seq(1,5)),
                 x2 = c(0.2*seq(1,5), 0.3*seq(1,5), 0.4*seq(1,5), 0.5*seq(1,5)),
                 a = rep(c(rep(1, 5), rep(2,5)), 2),
                 b = c(rep(1, 10), rep(2,10))
)

I now would like to have a facet_grid with x1 and x2 on top and a and b on the right where the values of a and b determine the line colour.

I tried

df.1 <- df %>%
  gather(x, xval, -t, -a, -b) %>%
  gather(p, pval, -t, -x, -xval) %>%
  distinct()

df.1$pval <- as.factor(df.1$pval)

ggplot(df.1, aes(t, xval)) +
  geom_line(aes(colour = pval)) +
  facet_grid(p~x)

and

dm.1 <- melt(df[, c("t", "x1", "x2")], id = 't')
colnames(dm.1) <- c("t", "x", "xval")
dm.2 <- melt(df[, c("t", "a", "b")], id = 't')
colnames(dm.2) <- c("t", "p", "pval")
dm <- merge(dm.1, dm.2)

dm$pval <- as.factor(dm$pval)
ggplot(dm, aes(t, xval)) +
  geom_line(aes(colour = pval)) +
  facet_grid(p~x)

But both do not give the desired result. Any hint would be greatly appreciated.

Edit: The desired result would be to have two lines in each facet similar to my first solution but the correct ones, i.e. straight lines and not the zig-zag lines that result.

Michael
  • 41,989
  • 11
  • 82
  • 128
Pascal
  • 563
  • 1
  • 3
  • 15

1 Answers1

0

The problem that's causing the zigzag plots you're getting is that there are multiple repeats of the same combinations of p and x, but there isn't a way to demarcate one from the other. So you get the first plot below:

library(tidyverse)

df <- data.frame(t = rep(seq(0,4), 4),
                                 x1 = c(1.2*seq(1,5), 1.3*seq(1,5), 1.4*seq(1,5), 1.5*seq(1,5)),
                                 x2 = c(0.2*seq(1,5), 0.3*seq(1,5), 0.4*seq(1,5), 0.5*seq(1,5)),
                                 a = rep(c(rep(1, 5), rep(2,5)), 2),
                                 b = c(rep(1, 10), rep(2,10))
)

df_long <- df %>%
    gather(key = x, value = xval, x1, x2) %>%
    gather(key = p, value = pval, a, b) %>%
    mutate(pval = as.factor(pval))

df_long %>%
    ggplot(aes(x = t, y = xval)) +
        geom_line(aes(color = pval)) +
        facet_grid(p ~ x)

You can see what it looks like here when I filter for a specific pair of x, p values. This will just place data at the same value of t repeatedly, instead of knowing how to make distinct lines.

df_long %>%
    filter(x == "x1", p == "a") %>%
    head()
#>   t  x xval p pval
#> 1 0 x1  1.2 a    1
#> 2 1 x1  2.4 a    1
#> 3 2 x1  3.6 a    1
#> 4 3 x1  4.8 a    1
#> 5 4 x1  6.0 a    1
#> 6 0 x1  1.3 a    2

Instead, before gathering, you can make an ID for each combination of a and b, and use that as your grouping variable in aes. There are probably other ways to do this, but a simple one is just interaction(a, b), which will give IDs that look like 1.1, 1.2, 2.1, 2.2, etc. Then add group = id inside your aes to make separate lines.

df_long_id <- df %>%
    mutate(id = interaction(a, b)) %>%
    gather(key = x, value = xval, x1, x2) %>%
    gather(key = p, value = pval, a, b) %>%
    mutate(pval = as.factor(pval))

df_long_id %>%
    ggplot(aes(x = t, y = xval, group = id)) +
    geom_line(aes(color = pval)) +
    facet_grid(p ~ x)

Created on 2018-05-09 by the reprex package (v0.2.0).

camille
  • 16,432
  • 18
  • 38
  • 60
  • Thanks! Maybe I didn't clarify enough. In facet x1,a there should only be the two lines where a is different but b is the same – Pascal May 09 '18 at 15:43
  • I think there's something you need to do differently in figuring out how to shape the data in that case, because when I filter your long-shaped data for (x1, a), there are 20 rows, i.e. enough to make 4 lines, not 2. – camille May 09 '18 at 15:59