Histogram set custom DisplayOrder

I created a histogram from data I extracted from a table. The input is as displayed in "size".

size = ["large" "xlarge" "small" "small" "medium" "large" "xlarge" "small" "medium" "medium" "xlarge"];

C = categorical(size);
h = histogram(C);

% right order would be:

    X = [4 1 2 3];
    Y = ["xlarge" "small" "medium" "large"];
    [Xsorted,I] = sort(X);
    Ysorted = Y(I);

%h.DisplayOrder = Ysorted;

This plot is what I get:

The command DisplayOrder only regognizes 'descend' or 'ascend'. As default the data is in order of the input data.

Is there a possibility to order the data according to Y?

Answer 1

[Steven Lord]

Specify the valueset in the order in which you want the categories to be listed when you construct the categorical array.

BTW, don't name a variable size. That already has a meaning in MATLAB.

sizes = ["large" "xlarge" "small" "small" "medium" "large" "xlarge" "small" "medium" "medium" "xlarge"];
Y = ["small" "medium" "large" "xlarge" ];

C = categorical(sizes, Y);
h = histogram(C);

Alternately use reordercats on the categorical array after the fact.

C2 = categorical(sizes);
categories(C2) % Note the order; look familiar from your original example?

ans = 4×1 cell array
    {'large' }
    {'medium'}
    {'small' }
    {'xlarge'}

C2 = reordercats(C2, Y);
categories(C2) % Note the new order

ans = 4×1 cell array
    {'small' }
    {'medium'}
    {'large' }
    {'xlarge'}

figure
histogram(C2)

[Katrin Niedermeier]

Thank you, that was easier than expected!

[Steven Lord]

By the way, since your data sounds like it has an inherent ordering (small < medium, medium < large, etc.) you might want to create it as ordinal data. Looking at a sample of your data:

sizes = ["large" "xlarge" "small" "small" "medium"];
Y = ["small" "medium" "large" "xlarge" ];
orderedC = categorical(sizes, Y, Ordinal=true)

orderedC = 1×5 categorical array
     large      xlarge      small      small      medium 

unorderedC = categorical(sizes, Y)

unorderedC = 1×5 categorical array
     large      xlarge      small      small      medium 

You could create a histogram with either:

subplot(1, 2, 1)
histogram(orderedC)
subplot(1, 2, 2)
histogram(unorderedC)

But the ordinal categorical lets you ask questions like:

orderedC(2) < orderedC(3) % Is xlarge < small? false

ans = logical
   0

orderedC(1) < orderedC(2) % Is large < xlarge? true

ans = logical
   1

You can't ask those same questions for the non-ordinal categorical.

unorderedC(2) < unorderedC(3)

Error using <
Relational comparisons are not allowed for categorical arrays that are not ordinal.