by Guangming Lang
~1 min read

Categories

  • r

Updated October 4, 2018

In an earlier post, I showed you how easy it is to make high quality histograms and boxplots using ezplot. When we’re analyzing a continuous variable, we often want to check if it’s normally distributed. Q-Q plot is a good tool to do that.

Make sure you first install ezplot by running the command devtools::install_github("gmlang/ezplot").

library(ezplot)
library(dplyr)

We’ll use the cars dataset, which comes with the base R distribution. It has two variables, speed and dist. Both are continuous. Let’s first create a normal Q-Q plot for dist.

plt = mk_qqplot(cars)
p = plt("dist", detrend = F)
square_fig(p)

center

We see dist is approximately normal because most of the data points are aligned linearly along the 45 degree diagonal line and within the confidence band. Next, we make a detrended normal Q-Q plot for speed and observe it’s also normally distributed because all data points are randomly scattered around $y = 0$ and within the confidence band.

plt("speed")

center

If you liked these how-to blog posts, you may want to check out my ezplot book.