Can you have a multilevel table?

I have a large table that has several groups of similar variables. I'd like to create levels of variables so I can group and access them easily- perhaps I need a struct or perhaps I am being lazy.

load patients
T1 = table(LastName,Gender,Age,Height,Weight,Systolic,Diastolic);
head(T1,3)

ans = 3×7 table
      LastName        Gender      Age    Height    Weight    Systolic    Diastolic
    ____________    __________    ___    ______    ______    ________    _________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    

I can merge variables, but I seem to lose the nested variable names.

T2 = mergevars(T1,{'Systolic','Diastolic'},'NewVariableName','BloodPressure');
head(T2,3)

ans = 3×6 table
      LastName        Gender      Age    Height    Weight    BloodPressure
    ____________    __________    ___    ______    ______    _____________

    {'Smith'   }    {'Male'  }    38       71       176       124     93  
    {'Johnson' }    {'Male'  }    43       69       163       109     77  
    {'Williams'}    {'Female'}    38       64       131       125     83  

I'd like to be able to access the Systolic and Diastolic variable under BloodPressure like this:

T2.BloodPressure.Systolic

%but not this

T2.BloodPressure(:,1)

It's a lot harder for me to keep track of the index, and I am hoping I would be able to use Tab Completion.

Also, in the event there is more than one 'Blood Pressure', say I have BloodPressure1 and BloodPressure2 both with sub variables Systolic and DIastolic, I'd like to be able to get all the Systolics at once- something like

T2.{:}.Systolic

Hope that explains what I'm after, and hoping there's a way to get there. Thanks.

Answer 1

[Voss]

You can use 'MergeAsTable',true in mergevars:

load patients
T1 = table(LastName,Gender,Age,Height,Weight,Systolic,Diastolic);
head(T1,3)

ans = 3×7 table
      LastName        Gender      Age    Height    Weight    Systolic    Diastolic
    ____________    __________    ___    ______    ______    ________    _________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    

T2 = mergevars(T1,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure', ...
    'MergeAsTable',true);
head(T2,3)

ans = 3×6 table
      LastName        Gender      Age    Height    Weight        BloodPressure    
                                                             Systolic    Diastolic
    ____________    __________    ___    ______    ______    _____________________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    

T2.BloodPressure

ans = 100×2 table
    Systolic    Diastolic
    ________    _________

      124          93    
      109          77    
      125          83    
      117          75    
      122          80    
      121          70    
      130          88    
      115          82    
      115          78    
      118          86    
      114          77    
      115          68    
      127          74    
      130          95    
      114          79    
      130          92    

T2.BloodPressure.Systolic

ans = 100×1

124 109 125 117 122 121 130 115 115 118

[Marcus Glover]

Thanks! Is there a way to grab all lower level variables that match? Like this:

load patients
T1 = table(LastName,Gender,Age,Height,Weight,Systolic,Diastolic);

T2 = head(mergevars(T1,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure', ...
    'MergeAsTable',true),3)

T2 = 3×6 table
      LastName        Gender      Age    Height    Weight        BloodPressure    
                                                             Systolic    Diastolic
    ____________    __________    ___    ______    ______    _____________________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    

T3 = renamevars(T2,{'Height','Weight'},{'Systolic','Diastolic'})

T3 = 3×6 table
      LastName        Gender      Age    Systolic    Diastolic        BloodPressure    
                                                                  Systolic    Diastolic
    ____________    __________    ___    ________    _________    _____________________

    {'Smith'   }    {'Male'  }    38        71          176         124          93    
    {'Johnson' }    {'Male'  }    43        69          163         109          77    
    {'Williams'}    {'Female'}    38        64          131         125          83    

T3=mergevars(T3,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure2', ...
    'MergeAsTable',true)

T3 = 3×5 table
      LastName        Gender      Age       BloodPressure2            BloodPressure    
                                         Systolic    Diastolic    Systolic    Diastolic
    ____________    __________    ___    _____________________    _____________________

    {'Smith'   }    {'Male'  }    38        71          176         124          93    
    {'Johnson' }    {'Male'  }    43        69          163         109          77    
    {'Williams'}    {'Female'}    38        64          131         125          83    

 
% I'd like to be able to access all Systolic, but can't figure it out:

T3.{'BloodPressure2','BloodPressure'}.Systolic
T3.{:}.Systolic

[Voss]

load patients
T1 = table(LastName,Gender,Age,Height,Weight,Systolic,Diastolic);

T2 = head(mergevars(T1,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure', ...
    'MergeAsTable',true),3);
T3 = renamevars(T2,{'Height','Weight'},{'Systolic','Diastolic'});

T3=mergevars(T3,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure2', ...
    'MergeAsTable',true)

T3 = 3×5 table
      LastName        Gender      Age       BloodPressure2            BloodPressure    
                                         Systolic    Diastolic    Systolic    Diastolic
    ____________    __________    ___    _____________________    _____________________

    {'Smith'   }    {'Male'  }    38        71          176         124          93    
    {'Johnson' }    {'Male'  }    43        69          163         109          77    
    {'Williams'}    {'Female'}    38        64          131         125          83    

 
% I'd like to be able to access all Systolic:
[T3.BloodPressure2.Systolic T3.BloodPressure.Systolic]

ans = 3×2

71 124 69 109 64 125

[Marcus Glover]

Thanks again. Is there any way to wildcard the 'BloodPressure' variable so that any and all Systolic subvariables are found?

[Seth Furman]

Starting in R2022a you can use patterns to index into tables (as well as in a number of table methods that accept table variable indices).

https://www.mathworks.com/help/matlab/ref/pattern.html

https://www.mathworks.com/help/matlab/release-notes.html?rntext=table+pattern&startrelease=R2022a&endrelease=R2022a

load patients
T1 = table(LastName,Gender,Age,Height,Weight,Systolic,Diastolic);
T2 = head(mergevars(T1,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure', ...
    'MergeAsTable',true),3);
T3 = renamevars(T2,{'Height','Weight'},{'Systolic','Diastolic'});
T3 = mergevars(T3,{'Systolic','Diastolic'}, ...
    'NewVariableName','BloodPressure2', ...
    'MergeAsTable',true)

T3 = 3×5 table
      LastName        Gender      Age       BloodPressure2            BloodPressure    
                                         Systolic    Diastolic    Systolic    Diastolic
    ____________    __________    ___    _____________________    _____________________

    {'Smith'   }    {'Male'  }    38        71          176         124          93    
    {'Johnson' }    {'Male'  }    43        69          163         109          77    
    {'Williams'}    {'Female'}    38        64          131         125          83    

bp = T3(:, "BloodPressure"+wildcardPattern)

bp = 3×2 table
       BloodPressure2            BloodPressure    
    Systolic    Diastolic    Systolic    Diastolic
    _____________________    _____________________

       71          176         124          93    
       69          163         109          77    
       64          131         125          83    

bpSystolic = varfun(@(t) t(:, "Systolic"), bp)

bpSystolic = 3×2 table
    Fun_BloodPressure2    Fun_BloodPressure
         Systolic             Systolic     
    __________________    _________________

            71                  124        
            69                  109        
            64                  125        

bpSystolic.Properties.VariableNames = bp.Properties.VariableNames

bpSystolic = 3×2 table
    BloodPressure2    BloodPressure
       Systolic         Systolic   
    ______________    _____________

          71              124      
          69              109      
          64              125      

Answer 2

[Adam Danz]

An alternative to @Voss's execellent answer is to create the nested table(s) and then add the nested table(s) as a table variable to the main table (MATLAB R2018b or later).

load patients
BloodPressure = table(Systolic,Diastolic); 
T1 = table(LastName,Gender,Age,Height,Weight, BloodPressure)

T1 = 100×6 table
      LastName        Gender      Age    Height    Weight        BloodPressure    
                                                             Systolic    Diastolic
    ____________    __________    ___    ______    ______    _____________________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    
    {'Jones'   }    {'Female'}    40       67       133        117          75    
    {'Brown'   }    {'Female'}    49       64       119        122          80    
    {'Davis'   }    {'Female'}    46       68       142        121          70    
    {'Miller'  }    {'Female'}    33       64       142        130          88    
    {'Wilson'  }    {'Male'  }    40       68       180        115          82    
    {'Moore'   }    {'Male'  }    28       68       183        115          78    
    {'Taylor'  }    {'Female'}    31       66       132        118          86    
    {'Anderson'}    {'Female'}    45       68       128        114          77    
    {'Thomas'  }    {'Female'}    42       66       137        115          68    
    {'Jackson' }    {'Male'  }    25       71       174        127          74    
    {'White'   }    {'Male'  }    39       72       202        130          95    
    {'Harris'  }    {'Female'}    36       65       129        114          79    
    {'Martin'  }    {'Male'  }    48       71       181        130          92    

[Jan Kappen]

I really like nested tables, but how can I run groupsummary on them? Without having the grouping functions supporting nested tables I don't see any advantages..

load patients
BloodPressure = table(Systolic,Diastolic); 
T1 = table(LastName,Gender,Age,Height,Weight, BloodPressure)

T1 = 100×6 table
      LastName        Gender      Age    Height    Weight        BloodPressure    
                                                             Systolic    Diastolic
    ____________    __________    ___    ______    ______    _____________________

    {'Smith'   }    {'Male'  }    38       71       176        124          93    
    {'Johnson' }    {'Male'  }    43       69       163        109          77    
    {'Williams'}    {'Female'}    38       64       131        125          83    
    {'Jones'   }    {'Female'}    40       67       133        117          75    
    {'Brown'   }    {'Female'}    49       64       119        122          80    
    {'Davis'   }    {'Female'}    46       68       142        121          70    
    {'Miller'  }    {'Female'}    33       64       142        130          88    
    {'Wilson'  }    {'Male'  }    40       68       180        115          82    
    {'Moore'   }    {'Male'  }    28       68       183        115          78    
    {'Taylor'  }    {'Female'}    31       66       132        118          86    
    {'Anderson'}    {'Female'}    45       68       128        114          77    
    {'Thomas'  }    {'Female'}    42       66       137        115          68    
    {'Jackson' }    {'Male'  }    25       71       174        127          74    
    {'White'   }    {'Male'  }    39       72       202        130          95    
    {'Harris'  }    {'Female'}    36       65       129        114          79    
    {'Martin'  }    {'Male'  }    48       71       181        130          92    

groupsummary(T1,"Gender",@(x)mean(x.Systolic),"BloodPressure")

Error using matlab.internal.math.checkDataVariables
Invalid data variables.

Error in groupsummary (line 247)
dataVars = matlab.internal.math.checkDataVariables(T, varargin{indStart}, 'groupsummary', 'Data');

[Adam Danz]

> I really like nested tables, but how can I run groupsummary on them?

@Jan Kappen, you can un-nest them to compute group summaries using splitvars.

Demo:

t3 = table(randi(2,5,1),rand(5,1),'VariableNames',{'a','b'});
t2 = table(rand(5,1),rand(5,1),t3,t3,'VariableNames',{'c','d','t3','t4'})

t2 = 5×4 table
       c          d             t3               t4      
                           a       b        a       b    
    _______    ________    _____________    _____________

    0.64143    0.087442    1      0.1543    1      0.1543
    0.10796     0.35111    2     0.86203    2     0.86203
    0.14392     0.91985    1     0.99177    1     0.99177
    0.15729     0.39593    2    0.085138    2    0.085138
    0.41819      0.1901    1      0.8832    1      0.8832

t1 = splitvars(t2)

t1 = 5×6 table
       c          d        t3_a      t3_b      t4_a      t4_b  
    _______    ________    ____    ________    ____    ________

    0.64143    0.087442     1        0.1543     1        0.1543
    0.10796     0.35111     2       0.86203     2       0.86203
    0.14392     0.91985     1       0.99177     1       0.99177
    0.15729     0.39593     2      0.085138     2      0.085138
    0.41819      0.1901     1        0.8832     1        0.8832

groupsummary(t1,'t3_a', 'mean', 't3_b')

ans = 2×3 table
    t3_a    GroupCount    mean_t3_b
    ____    __________    _________

     1          3          0.67642 
     2          2          0.47358 

[Jan Kappen]

@Adam Danz that's a very nice information, thx.