Multigroup Boxplot dimension error

Hello, I am struggling with a Boxplot using multigroup. My data is formatted as below. I like to apply a grouping variable for the columns and also for the rows. In this case, the 3 values per column belong to an observed, simulated and expected value. Within the rows, there are different settings. This is a minimal working example, but I really don't understand what the error is about, because all rows/columns match from my point of view.

But I get the error:

Error using boxplot>convertToCellarrayOfColumnVectors (line 1082)
All columns in 'G' must have the same length.

data = rand(3,10);
groupH = [0,1,0,3,1,3,0,1,1,0];
groupV = repmat([0;1;2],1,10);
boxplot(data, {groupH, group})

The Output as desired as having 3 Boxplots for same category out of h with 3 different classes from groupV directly beside. I found the following example after posting, which is quite similar, but I am really confused adapting this code for my purposes.

Desired Output

[KALYAN ACHARJYA]

Can you share the code/representational code??

[jonas]

How do you call Boxplot?

[Jan Maluche]

Sorry, I don't know why I forget to post the line. Edited the code right now.

[jonas]

Can you explain what your desired output is? How many boxes are you looking for?

[Jan Maluche]

Updated the question with the desired output. For this case I am expecting 30 boxes.

[jonas]

Okay I see the problem. Will post an answer shortly.

Answer 1

[jonas]

I found some issues with your approach. Most importantly, the number of rows in the matrix is supposed to represent the number of data points per series, which are then later visualized by a single box. Perhaps you understood this already, but each series in your example had only a single point, which was confusing to me. I have changed this to 100 pts per series of data.

This example goes on to show how to group 30 series of data in 10 different categories [1st 1st 1st 2nd 2nd 2nd ... 10th 10th 10th], with each category having three different types of series [1st 2st 3rd].

% Some data
data = rand(100,30); % 30 series, 100 pts in each

% 10 categories
% a trick to get [0 0 0 1 1 1 2 2 2 ... 9 9 9]
groupH = repmat([0:9],1,3); 
groupH = reshape(groupH,10,3)'
groupH = groupH(:)

% three types of data series
groupV = repmat([0 1 2],1,10); %every third series is of the same type

% plot 
boxplot(data,{groupH,groupV},'factorgap',[5 2])

.

You can see the two grouping vectors in the xlabels. Note that the grouping [0 1 2...] is redundant in this example. However, if your data is not nicely organized, then two grouping variables can be useful.

[Jan Maluche]

Thank you, this is really helpful and totally adresses my problem.

[jonas]

Happy to help! Please mark the answer as accepted if your issue is resolved :)