Scatter plot in R with ggvis: how to plot multiple groups with different shape markers and corresponding fitted regression lines

Question

To plot the following in R with the ggvispackage,

enter image description here

the code is

mtcars %>% 
  ggvis(~wt, ~mpg, fill = ~factor(cyl)) %>% 
  layer_points() %>% 
  group_by(cyl) %>% 
  layer_model_predictions(model = "lm")

If I change the fill to shape in the above, there would be an error:

Error: Unknown properties: shape.
Did you mean: stroke?

Why? How to achieve the desired outcome?

If you don't use `group_by` and `layer_model_predictions`, ggvis can take shape. So, I am guessing, when you run lm for each level of cyl, something is going on. — jazzurro, Aug 31 '14 at 11:50

Andrie · Accepted Answer · 2014-08-31T17:23:48.293

6

You have to specify the shape in the layer_points() call:

mtcars %>% 
  transform(cyl = factor(cyl)) %>%
  ggvis(~wt, ~mpg) %>% 
  layer_points(shape = ~cyl, fill = ~cyl) %>% 
  group_by(cyl) %>% 
  layer_model_predictions(model = "lm", formula=mpg~wt)

(Note that I use transform() to convert cyl into a factor. This means you don't have to convert cyl into a factor in the ggvis() call, and the plot key is a little bit neater.)

enter image description here

edited Aug 31 '14 at 17:23

answered Aug 31 '14 at 17:18

Andrie

176,377
47
447
496

In the `dplyr`/`ggvis` spirit, I would use `mutate()` over `transform()` here. `mutate(cyl = factor(cyl)) %>%` – Steven Beaupré Dec 18 '14 at 13:58

Scatter plot in R with ggvis: how to plot multiple groups with different shape markers and corresponding fitted regression lines

1 Answers1