1
$\begingroup$

I'm a web developer and currently working on visualizing the execution time of processing a request. Thus, I can create a list with all execution times (e.g., t = [0.12, 0.23, 0.16, 0.17, ...]). The problem is that the size of the list can grow enormously, for example: length(t) = 100,000 is not unusual. Now, the plotting program that I use for making boxplots becomes quite slow with 100k datapoints.

Would it be possible to generate the same boxplot using the following 5 datapoints: [min(t), q1(t), median(t), q3(t), max(t)]? If not, is there any other way to simplify the boxplot constructed from t?

$\endgroup$
  • $\begingroup$ In addition to asking if it’s possible, you may want to reflect on how you’re using the visualization. Is the min and max useful information? Is the 95th percentile more useful? What about the mean? $\endgroup$ – kbrose Apr 26 '18 at 13:56
1
$\begingroup$

There is no problem with this. There are different types of boxplots. The most common is to extend the 'whiskers' out to the furthest points that are not further than 1.5 times the inter-quartile range above the third quartile and below the 1st quartile. All remaining points beyond that are plotted individually as potential outliers. Another way is to just have the whiskers go all the way to the minimum and maximum values. You are referring to the latter.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.