Odd `mark_density()`

[Earo Wang]

penguins %>% 
  vega(enc(x = body_mass_g, colour = species)) %>% 
  mark_density()

[Stuart Lee]

Yeah this one definitely doesn't work how I expected / is broken try:

penguins %>% 
  vega(enc(x = body_mass_g, colour = species)) %>% 
  mark_density()

[Stuart Lee]

Missing values causing issues again.
This gives the expected result:

penguins %>% 
    filter(!is.na(body_mass_g)) %>% 
    vega(enc(x = body_mass_g, colour = species)) %>% 
    mark_density()
    
 penguins %>% 
    filter(!is.na(body_mass_g)) %>% 
    vega(enc(x = body_mass_g)) %>% 
    mark_density() 

[Earo Wang]

😢 Now this mtcars not working as expected.

mtcars %>%
  mutate(cyl = factor(cyl)) %>%
  vega(encoding = enc(x = wt, colour = cyl)) %>%
  mark_density()

[Stuart Lee]

it seems like an issue with the stacking again? here's ggplot2 for reference

[Earo Wang]

I've removed "stack" = zero". The default in ggplot2 is "identity". In vegalite, stack = zero means stacked density?

[Earo Wang]

Seems not working for mtcars when stacking.

[Stuart Lee]

maybe the default bandwidth is off? maybe this more reason for us to compute these things on the r side haha

[Stuart Lee]

Should position = "stack" be the default?