"Data variables must be numeric or logical" despite variable being logical when using grpstats

function ViewData(data)
resp = cell2mat(data(:, 3)) == 39;
corrAns = ismember(data(:, 1), {'2F_H', '2F_V'});
score = double(corrAns == resp);
ifi = cell2mat(data(:, 2));
orient2 = ismember(data(:,1), {'1F_H', '2F_H'});

tblData = table(score, ifi, data(:, 1), orient2, 'variablenames', {'score', 'ifi', 'trialType', 'orient2'});
% tblData = table(score, ifi, orient, data(:, 1), 'variablenames', {'score', 'ifi', 'trialType'});
tblMean = grpstats(tblData, {'orient2', 'ifi'}); %change trialType to orient, or vice versa

disp(tblMean);
end

So the error is "Error using dsgrpstats (line 293) Data variables must be numeric or logical."

But I'm confused because orient2 is literally a logical, no? If I define tblMean in this way, however, it works:

tblData = table(score, ifi, data(:, 1), orient2, 'variablenames', {'score', 'ifi', 'trialType', 'orient2'});
% tblData = table(score, ifi, orient, data(:, 1), 'variablenames', {'score', 'ifi', 'trialType'});
tblMean = grpstats(tblData, {'trialType','orient2', 'ifi'}); 

and the column for orient2 LITERALLY is "true" or "false" all the way down. AND when I do class(tblData.trialType), it's "cell" which adds even more to the confusion.

I'm sure it's a simple fix but I can't see it. Thank you!

Answer 1

[Cris LaPierre]

The issue is not your grouping variable, it is your tblData. Specifically, trialType, which is a cell, and not numeric or logical. Here's a simple example using a built-in data set with your names.

load('patients.mat')
orient2 = ismember(Gender,{'Male'});
tblData = table(Age,Diastolic,Systolic,orient2,'VariableNames',{'score', 'ifi', 'trialType', 'orient2'});

% All table variables are numeric or logical
tblMean = grpstats(tblData, {'orient2', 'ifi'})

tblMean = 41×5 table
            orient2    ifi    GroupCount    mean_score    mean_trialType
            _______    ___    __________    __________    ______________

    0_68     false     68         1               42             115    
    0_70     false     70         1               46             121    
    0_73     false     73         2               38             122    
    0_74     false     74         2             39.5             120    
    0_75     false     75         3               41          119.67    
    0_76     false     76         3           30.667          121.67    
    0_77     false     77         3               37          117.33    
    0_78     false     78         1               41             117    
    0_79     false     79         7           37.714          119.43    
    0_80     false     80         6           33.833           119.5    
    0_81     false     81         2             37.5           117.5    
    0_82     false     82         2               40             124    
    0_83     false     83         3               42          124.33    
    0_85     false     85         3           32.667          120.67    
    0_86     false     86         2               39           117.5    
    0_87     false     87         1               31             123    

% Table variable Gender (renamed to trialType) is a cell
tblData = table(Age,Diastolic,Gender,orient2,'VariableNames',{'score', 'ifi', 'trialType', 'orient2'});
tblMean = grpstats(tblData, {'orient2', 'ifi'})

Error using dsgrpstats (line 293)
Data variables must be numeric or logical.

Error in grpstats (line 144)
[varargout{1:nargout}] = dsgrpstats(x,group,whichstats,varargin{:});

Use the 'DataVars' name-value pair in the input to specify which numeric or logical variables to compute stats for.

Your second approach works because the cell data is now a grouping variable, so stats are not calculated for it. Perhaps you want to do something like this?

function ViewData(data)

resp = cell2mat(data(:, 3)) == 39;

corrAns = ismember(data(:, 1), {'2F_H', '2F_V'});

score = double(corrAns == resp);

ifi = cell2mat(data(:, 2));

orient2 = ismember(data(:,1), {'1F_H', '2F_H'});

tblData = table(score, ifi, data(:, 1), orient2, 'variablenames', {'score', 'ifi', 'trialType', 'orient2'});

tblMean = grpstats(tblData, {'orient2', 'ifi'},'mean','DataVars','score')

end

[Gabriel Wooten-Soto]

Thank you so much! so in my second attempt, how come trialType functions as a grouping variable and orient2 doesn't? Is it because it's a cell?

[Cris LaPierre]

In your second attempt, you have specified 3 grouping variables. All three are used. The data is first grouped by 'trialType', then, within each of those groups, by 'orient2', and finally within each of those groups, by 'ifi'.

Using the same example from above, you can see that in the output below.

1. The data is first grouped by 'trialType'. {'Male'} first, them {'Female'}
2. Within each of those groups, the data is grouped by 'true' and 'false'
3. Within each of those groups, the data is grouped by 'ifi'.

The GroupCount column shows you how many items are in the specified grouping combination. So in the first row, for the group trialType=Male, orient2=true and ifi=74 there were 3 scores averaged together.

load('patients.mat')
orient2 = ismember(Gender,{'Male'});
tblData = table(Age,Diastolic,Systolic,orient2,'VariableNames',{'score', 'ifi', 'trialType', 'orient2'});

% Table variable Gender (renamed to trialType) is a cell
tblData = table(Age,Diastolic,Gender,orient2,'VariableNames',{'score', 'ifi', 'trialType', 'orient2'});
tblMean = grpstats(tblData, {'trialType','orient2', 'ifi'});
head(tblMean)

ans = 8×5 table
                 trialType    orient2    ifi    GroupCount    mean_score
                 _________    _______    ___    __________    __________

    Male_1_74    {'Male'}      true      74         3           37.667  
    Male_1_75    {'Male'}      true      75         1               25  
    Male_1_76    {'Male'}      true      76         2               43  
    Male_1_77    {'Male'}      true      77         4             37.5  
    Male_1_78    {'Male'}      true      78         3               34  
    Male_1_79    {'Male'}      true      79         2             43.5  
    Male_1_80    {'Male'}      true      80         2               40  
    Male_1_81    {'Male'}      true      81         2               36  

tail(tblMean)

ans = 8×5 table
                   trialType     orient2    ifi    GroupCount    mean_score
                   __________    _______    ___    __________    __________

    Female_0_85    {'Female'}     false     85         3           32.667  
    Female_0_86    {'Female'}     false     86         2               39  
    Female_0_87    {'Female'}     false     87         1               31  
    Female_0_88    {'Female'}     false     88         2               37  
    Female_0_90    {'Female'}     false     90         1               44  
    Female_0_91    {'Female'}     false     91         4            39.25  
    Female_0_92    {'Female'}     false     92         2               41  
    Female_0_96    {'Female'}     false     96         2             38.5