Comment Permalink
140
The graphs would be more readable if the log1p transformation applied to the coordinates, but not to the labels.
qplot(Age,Charity,color=EffectiveAltruism,data=survey)+geom_point(size=I(3))+coord_trans(y="log1p")
Unfortunately the choice of labels is not great.
100
Thanks, I didn't know ggplot2 had a feature like that. Some more googling reveals you can override the ticks on the axis by specifying your own breakpoints on the y-axis, so one can write:
qplot(Age, Charity, color=EffectiveAltruism, data=survey) +
geom_point(size=I(3)) +
scale_y_continuous(breaks=round(exp(1:10))) + coord_trans(y="log1p")
which yields more labels for the lower range of donations:
http://www.gwern.net/EA%20donations