Boxplot Data

I have a very large set of data with many groups and subgroups. The boxplot command works well for visualization of the data. I was wondering if there was an easy way to extract the data displayed without actually doing a manual calculation of each parameter. For example, I wish boxplot provided a set of function output variables that report the values used to plot each box (mean, interquartile range, outliers, etc.)

Does anyone know of an easy way to extract this information automatically?

I know that I can manually write a script to calculate each of the values...just wondering if there is a shortcut.

Answer 1

[Maria Battle]

Similar to Oleg's answer, you could use the functions iqr and mean to address your first concerns. Note: interquartile uses median, not mean. For interquartile outlier output, use the 'quartiles' method for isoutlier. To get the same results between boxplot and isoutlier, make sure the whisker value in boxplot is the same value as the threshold factor in isoutlier.

load examgrades 
t = datetime('2015-12-18 12:00:00') + days(1:120)';
g = grades(:,1); 

figure
boxplot(g, 'Whisker', .5); 

io = isoutlier(g,'quartiles','ThresholdFactor', .5);
outg = g(io);
outt = t(io);
t = table(outt, outg);

[Aymen ABID]

isoutlier not exist in r2016

[Maria Battle]

In that case, you might try using Matlab's quantile function and create a condition for your threshold. I believe that's basically what isoutlier() is doing when you use the quartiles and ThresholdFactor options. I haven't tried it though.

Answer 2

[Oleg Komarov]

It way easier to calculate the stats:

min
max
quantile(x,[.25 .5 .75])
mean

[Rania Rayyes]

median and not mean, and the upper bound is not max, max is outlier here

[Aymen ABID]

@Rania Rayyes quabtile .25 .5 .75 gives Q1,Q2 (median),Q3

median with isnan give the median

Answer 3

[zahra zol]

i am desperately looking for the same thing...

[Samuel Geurts]

if you open de handle of the boxplot you can achieve that information, but that is not easier than calculating it.

 h = boxplot(x)
 handle(h)