Boxchart function gives empty group

Hi,

I'm trying to make a boxplot for my data. I made a table with my grouping variable (T.Group) and a value that I want to plot (T.CorrVal) (see attachment). I use the following code:

load T
b = boxchart(T.Group, T.CorrVal);

This gives me boxplots for my two groups (CaCl and Sham), but also an extra group called empty. I thought this was because I have some NaN values in my T.CorrVal, but the weird thing is that if I plot the exact same thing with another grouping variable (4 groups: CaClFemale, CaClMale, ShamFemale, ShamMale), like this:

b = boxchart(T.Combi, T.CorrVal);

it does give me the right boxplots, without an extra empty one.

How do I fix this? Thank you in advance

Answer 1

[Cris LaPierre]

This is because your data is categorical. For this datatype, the categories are independent (in some respect) of the values. boxchart is just creating a plot for each category. Your variable just happens to have no 'empty' values, so the plot is showing zero. You can use categories to see what categories the variable accepts.

Before creating the boxchart, try removing all empty categories from your categorical variable using removecats.

% Create a dummy data set
Group = categorical(["CaCl";"Sham";"empty"]);

% Fill Group with values using 2 of the category names
ind = randi(2,100,1);
Group = Group(ind);

CorrVal = rand(size(Group));
T = table(Group,CorrVal)

T = 100×2 table
    Group    CorrVal 
    _____    ________

    CaCl      0.40479
    Sham      0.90618
    Sham      0.44116
    CaCl      0.58259
    CaCl      0.42117
    Sham      0.30017
    CaCl      0.94303
    Sham      0.38962
    Sham      0.70232
    CaCl      0.25015
    Sham      0.33066
    Sham     0.045428
    CaCl      0.93625
    CaCl       0.8541
    CaCl      0.61066
    CaCl      0.97252

% View the number of each group in T
groupsummary(T,"Group",'IncludeEmpty',true)

ans = 3×2 table
    Group    GroupCount
    _____    __________

    CaCl         50    
    Sham         50    
    empty         0    

% View the category names in T.Group. Note that empty appears, even though
% there are 0 occurrences.
categories(T.Group)

ans = 3×1 cell array
    {'CaCl' }
    {'Sham' }
    {'empty'}

% Create a boxchart. empty appears here, too.
boxchart(T.Group,T.CorrVal)

% Now remove empty categories from T.Group and repeat
T.Group = removecats(T.Group);
groupsummary(T,"Group",'IncludeEmpty',true)

ans = 2×2 table
    Group    GroupCount
    _____    __________

    CaCl         50    
    Sham         50    

categories(T.Group)

ans = 2×1 cell array
    {'CaCl'}
    {'Sham'}

figure
boxchart(T.Group,T.CorrVal)

[Marleen]

Thank you so much! It works.

[Cris LaPierre]

Sorry, missed that you uploaded your data. Here are the steps to use to inspect the categories and then remove the unused category name from your variable.

load T
categories(T.Group)

ans = 3×1 cell array
    {'CaCl' }
    {'Sham' }
    {'empty'}

T.Group = removecats(T.Group);
categories(T.Group)

ans = 2×1 cell array
    {'CaCl'}
    {'Sham'}

b = boxchart(T.Group, T.CorrVal);