CODEWITHSUNDEEP
rggplot2
Robbo
Robbo

Reputation: 3

Edit stat_smooth lines AND geom_points in the same figure

I have a graph with points from multiple levels of a categorical variable and the relevant loess smoothing lines for each level of the same variable. I need to edit BOTH the points AND the lines as in order convert the default colour graph to a meaningful black and white graph. I can edit the points using scale_shape_manual but not the lines. The following code uses mtcars to get close to what I need. Unfortunately all the liens are the same color (red)I suspect I don't understand where these sorts of ?aesthetic? commands should be placed. Any assistance greatly appreciated.

c <- ggplot(mtcars, aes(y=wt, x=mpg, shape=factor(cyl)))
c + 
  stat_smooth(method=loess, size = 1, col="red") + 
  geom_point(aes(fill = factor(cyl)), size = 4) + 
  scale_shape_manual(values=c(16,22, 1))

ggplot produced by this code

Upvotes: 0

Views: 445

Answers (1)

chemdork123
chemdork123

Reputation: 13803

There are a few ways select specific colors for discrete and gradient scales in ggplot2, mostly centering around usage of the scale_color_... and scale_fill_... functions here in a similar way to how you have used scale_shape_manual().

Since you're looking for grayscale, it's probably most straightforward to having those scales selected automatically for you using the convenience functions scale_color_grey() and scale_fill_grey().

Finally, in order to have ggplot2 select the color of each line, you need to have the legend created for that and place the color= aesthetic within aes(). You had col= (another way to indicate color=) outside of aes() for stat_smooth(), so all lines were drawn with that color. I believe this is what you are looking to do:

ggplot(mtcars, aes(y=wt, x=mpg, shape=factor(cyl))) +
  stat_smooth(method=loess, size = 1, aes(color=factor(cyl))) +
  geom_point(aes(fill = factor(cyl)), size = 4) +
  scale_shape_manual(values=c(16,22, 1)) +
  scale_color_grey() + scale_fill_grey()

enter image description here

You'll notice it's hard to see the lightest line color. You can fix that by applying a bit of transparency to the fill around geom_smooth() via alpha= and also changing to the theme_bw() to have a white background:

ggplot(mtcars, aes(y=wt, x=mpg, shape=factor(cyl))) + theme_bw() +
  stat_smooth(method=loess, size = 1, aes(color=factor(cyl)), alpha=0.18) +
  geom_point(aes(fill = factor(cyl)), size = 4) +
  scale_shape_manual(values=c(16,22, 1)) +
  scale_color_grey() + scale_fill_grey()

enter image description here

Upvotes: 1

Related Questions