There is another very useful kind of regression, quantiles. Drawing them under the ggplot2
package it's not challenging; it has a whole quantile dedicated function, geom_quantile()
. Drawing them using ggvis
and plotly
is also possible, but demands way more code.
This recipe draws 20 percent, 40 percent, 60 percent, and 80 percent quantile regression lines in a diamonds' carat versus price scatterplot. Drawing are amde using ggplot2
and plotly
respectively.
Quantile regressions come from the quantreg
package; make sure to have it installed:
> if( !require(quantreg)){ install.packages('quantreg')}
Even ggplot2
relies on quantreg
to fit quantile regressions.