how to concatenate tables stored in a structure

I have a structure where tables with same variable names are stored. I want to concatenate all these tables in one big table.

say i have struct S

struct with fields:

      T1: [34113×16 table]
      T2: [34133×16 table]
      T3: [34059×16 table]
      T4: [33297×16 table]
      T5: [34150×16 table]

I can do:

T=cat(1,T1,T2,T3,T4,T5)

but want to do it with listing struct fields and concatenate.

[Stephen23]

This would be a lot simpler with a non-scalar structure:

S(1).T = ...
S(2).T = ...
S(3).T = ...
...

then all you would need to do is:

cat(1,S.T)

Putting numbers into fieldnames just makes everything more complicated.

Answer 1

[Guillaume]

As Stephen commented, the initial designed is flawed. numbered variables and field names should always be replaced by a single container.

With a structure it's easily fixed by converting the structure into a cell array:

C = struct2cell(S);
vertcat(C{:})   %vertically concatenate all the fields of S

[Jan]

+1. If the data are represented as fields of a struct, this is the easiest solution. Stephen's comment showed, how to avoid a list of "numbered" fields by using a struct array. Alternatively you could store the tables in a cell array also: T{1}, T{2}, ... or if you need it: S.T{1}, S.T{2}, ... Then the concatenation is done again by:

vertcat(S.T{:})   % or vertcat(T{:})

Answer 2

[Peter Perkins]

If you really ever only have five fields/tables in your scalar struct, just do this:

T=cat(1,S.T1,S.T2,S.T3,S.T4,S.T5)

But presumably you already knew that and you have either a large number of fields/tables, or a variable number. Try this (which uses numeric scalars instead of tables, but it's exactly the same):

>> s = struct; s.A = 1; s.B = 2; s.C = 3
s = 
  struct with fields:

    A: 1
    B: 2
    C: 3
>> c = struct2cell(s)
c =
  3×1 cell array
    {[1]}
    {[2]}
    {[3]}
>> vertcat(c{:})
ans =
     1
     2
     3

Answer 3

[John BG]

Hi effess

this is John BG <mailto:[e-mail] [e-mail]>

there's no need to use the command cat or vertcat.

just concatenate the tables directly, like in the following example

1.

generating example data

var1=[1 2 3];var2=[4 5 6]
T1=table(var1,var2)
T2=T1
T3=T1
S(1).T=T1
S(2).T=T2
S(3).T=T3

2.

Since you clearly mentioned that all tables have the same variables, just put them all together like this:

T1=[S(1).T;S(2).T;S(3).T]
=

    3×2 table
         var1           var2    
      ___________    ___________
      1    2    3    4    5    6
      1    2    3    4    5    6
      1    2    3    4    5    6

.

effess

if you find this answer useful would you please be so kind to consider marking my answer as Accepted Answer?

To any other reader, if you find this answer useful please consider clicking on the thumbs-up vote link

thanks in advance for time and attention

John BG

<mailto:[e-mail] [e-mail]>

[Guillaume]

there's no need to use the command cat or vertcat

Note that [a, b] is just a short hand for horzcat. When you write [a, b], matlab ends up invoking horzcat(a, b).

[John BG]

same variables, the tables have the same variables. Of course you can do many different things to put together tables, but I show the simple way that is easiest to use when: both tables have exactly the same variables.

[Jan]

John, you wrote "there's no need to use the command cat or vertcat" and suggested [S(1).T;S(2).T;S(3).T]. But [;] and vertcat are completely equivalent and just a different notation.

Stephen's suggestion cat(1, S.T) is easier and does not depend on the number of fields.

[John BG]

To effess

Actually the generation of the structure as well as the concatenation can be simplified and automated with the powerful and well known command evalin:

Example

.

var1=[1 2 3];var2=[4 5 6]
T1=table(var1,var2)
T2=T1
T3=T1

var1=var1*10;var2=var2*11
T10=table(var1,var2)
table_names={'T1' 'T2' 'T3' 'T10'}

1.- structure storage

str_struct=[]
for k=1:1:size(table_names,2)
    str_struct=[str_struct 'S(' num2str(k) ').T=' table_names(k) ';']
end
str_struct=cell2mat(str_struct)

evalin('base',str_struct)

.

check

.

S(1).T
S(2).T
S(3).T
S(4).T

    =
     1×2 table
          var1           var2    
       ___________    ___________
       1    2    3    4    5    6
    =
     1×2 table
          var1           var2    
       ___________    ___________
       1    2    3    4    5    6
    =
     1×2 table
          var1           var2    
       ___________    ___________
       1    2    3    4    5    6
    =
     1×2 table
             var1                 var2       
       _________________    _________________
       100    200    300    484    605    726

.

2.- Concatenation

.

str1=[];
for k=1:1:size(table_names,2) 
    str1=[str1 table_names{k} ';']; 
end

str1(end)=[]
str1=['Tresult=' '[' str1 ']']

evalin('base',str1)

.

check

Tresult
  =
   4×2 table
           var1                 var2       
     _________________    _________________
       1      2      3      4      5      6
       1      2      3      4      5      6
       1      2      3      4      5      6
     100    200    300    484    605    726

.

effess

if you find this answer useful would you please be so kind to consider marking my answer as Accepted Answer?

To any other reader, if you find this answer useful please consider clicking on the thumbs-up vote link

thanks in advance for time and attention

John BG

<mailto:[e-mail] [e-mail]>

[John BG]

and when the tables have exactly the same variables, just in case there was any doubt, there's no need to use vertcat, or horzcat, or any cat of any type, the more compact semi colon notation works ok.

Regards

John BG

<mailto:[e-mail] [e-mail]>

[Jan]

Suggesting evalin('base') to create a struct array (in the base workspace!) is complete useless and a bad programming practice. Compare this with the 2 lines of code suggested by Guillaume, which are working inside a function also.

John, you insist on claiming that horzcat is not needed and do use it repeatedly in form of its abbreviation [;] . This is a slapstick comedy.

[Jan]

@Rene Ferdinands: I've read your (meanwhile deleted) opinion about my "condescending" replies. Take into account, that this discussion has a background you are not aware of. Many contributions have been removed by the admins due to rudeness, so if you can see only my parts of the dicsussion, does not necessarily mean, that I was the aggressive part.

I appreciate that you care about the polite and friendly tone in the forum.