stat_cor() limitation in GGPLOT2
vmatli opened this issue · comments
Please briefly describe your problem and what output you expect.
I was using stat_cor() to report correlation between two sets of values in GGPLOT. There was a skew in the data so I used scale_x_continuous and scale_y_continuous to log transform the values. When I used the stat_cor() function it was providing values on transformed scale rather than untransformed scale. I understand that having a different R would cause confusion due to the pattern observed in the data but having the option to use transformed vs untransformed values could be beneficial.
Random Data
x<-rnorm(20000,25,4)
y<-rnorm(20000,17,2)
Make Dataframe
data<-data.frame(cbind(x,y))
Expected behavior
Untransformed scale plot
plot<-ggplot(data,aes(x,y))+
geom_point()+
stat_cor()
plot
Actual behavior
Transformed scale plot
plot_trans<-ggplot(data,aes(x,y))+
geom_point()+
stat_cor()+
scale_x_continuous(trans="log")+
scale_y_continuous(trans="log")
plot_trans
Steps to reproduce the problem
If there was an option in stat_cor to show r values on transformed scale vs untransformed scale it would make the whole process more efficient. I dont know how to go about fixing the issue so maybe someone can pitch in.
Thanks