ggplot2中的stat_density2d（）问题

Question

My dataset (origen) is composed by 3 columns, a factor variable (Origen), and two cuantitative variables (Ae and Adp). Some factors of the origen categorical variable are composed by only 1 or 2 observations.

When running this code:

ggplot (origen, aes (x = Ae, y = Adp, colour = Origen)) + 
stat_density2d(aes (fill = Origen, colour = Origen, alpha = 0.9),
geom = "polygon", lty = 3,lwd = 0.001) + xlim(8, 23) + ylim(22, 58) + 
scale_alpha(range = c(0, 0.2),guide = "none") + 
geom_point(aes(fill=Origen),colour="black",pch=21,size=3,alpha=0.1)

I get no results for the stat_density2d function within the ggplot() one. However, when I manually remove those groups which contain only 1 or 2 observations at most, the above code works properly and the density areas appear.

How can I avoid this issue by modifying the code? What new attributes must be writen in the stat_density2d function to run those more-than-two-observations groups as well as the points of the other groups?

Answer 1

This is not the direct answer to your question, but you can remove the group with <1 observations by doing group_by() and filter() .

library(dplyr)

data_filtered <- data %>%
  group_by(Origen) %>%
  filter(n() > 1)

(The warning "missing value where TRUE/FALSE needed" seems to be raised from here: https://github.com/tidyverse/ggplot2/blob/8778b48b37d8b7e41c0f4f213031fb47810e70aa/R/stat-density-2d.r#L58 )