I'm creating a facetted plot to view predicted vs. actual values side by side with a plot of predicted value vs. residuals. I'll be using shiny to help explore the results of modeling efforts using different training parameters. I train the model with 85% of the data, test on the remaining 15%, and repeat this 5 times, collecting actual/predicted values each time. After calculating the residuals, my data.frame looks like this:

head(results)
       act     pred       resid
2 52.81000 52.86750 -0.05750133
3 44.46000 42.76825  1.69175252
4 54.58667 49.00482  5.58184181
5 36.23333 35.52386  0.70947731
6 53.22667 48.79429  4.43237981
7 41.72333 41.57504  0.14829173

What I want:

  • Side by side plot of pred vs. act and pred vs. resid
  • The x/y range/limits for pred vs. act to be the same, ideally from min(min(results$act), min(results$pred)) to max(max(results$act), max(results$pred))
  • The x/y range/limits for pred vs. resid not to be affected by what I do to the actual vs. predicted plot. Plotting for x over only the predicted values and y over only the residual range is fine.

In order to view both plots side by side, I melt the data:

library(reshape2)
plot <- melt(results, id.vars = "pred")

Now plot:

library(ggplot2)
p <- ggplot(plot, aes(x = pred, y = value)) + geom_point(size = 2.5) + theme_bw()
p <- p + facet_wrap(~variable, scales = "free")

print(p)

That's pretty close to what I want:

enter image description here

What I'd like is for the x and y ranges for actual vs. predicted to be the same, but I'm not sure how to specify that, and I don't

r ggplot2 facet
Edit
Report