CODEWITHSUNDEEP
rprobability-densitydensity-plot
elcortegano
elcortegano

Reputation: 2694

Is it possible to know the maximum value of density in advance?

Consider the following example plot:

mtcars %>% ggplot() + stat_bin(aes(x=wt, y=..density..), binwidth=0.2)

The maximum value of density is near to 0.8. Is it possible to know its exact value?

According to stat_bin documentation, ..density.. represents the density of points in bin, scaled to integrate to 1, but this does not help me to much to understand how these values are computed in practice, or how to get the maximum density point (as it also depends on the binwidth set).

The closer I have get is with:

binwidth = 0.2
((mtcars$wt %/% binwidth)*binwidth) %>% table %>% prop.table %>% max

But it does return the value observed in the plot.

Any ideas?

Upvotes: 1

Views: 1064

Answers (2)

Jordi Puigdellivol
Jordi Puigdellivol

Reputation: 76

you can try this

binwidth=0.2
x=mtcars$wt
x_bins <- table(cut(x,breaks=seq(min(x),max(x),by = binwidth)),useNA = "ifany" )
max(x_bins)/(sum(x_bins)*binwidth )
#[1] 0.78125

or 

gg_plot <- mtcars %>% ggplot() + stat_bin(aes(x=wt, y=..density..), binwidth=binwidth)
max(ggplot_build(gg_plot)$data[[1]][[1]])

Upvotes: 1

G. Grothendieck
G. Grothendieck

Reputation: 270338

Not clear what in advance refers to but if you are looking for the maximum of the density curve then try this:

dens <- density(mtcars$wt)
plot(dens)

max(dens$y)
## [1] 0.5068726

Upvotes: 2

Related Questions