How to split a column with numerical values?

Hi all,

I have a 1x72 structure with one column containing three numerical values (attched).

I would like to convert it into a 3x72 table with each numerical value in a seperate column with diffrent name, for example:

'name 1' 'name 2' 'name 3'

0.9377 0.9246 0.9982

Can you help please?

Answer 1

[Stephen23]

Here are two robust MATLAB approaches (both much more robust than blindly using STRUCT2ARRAY):

S = load('data.mat');
ICCs = S.ICCs

ICCs = 1×72 struct array with fields:
    Metrics

T = array2table(vertcat(ICCs.Metrics),'VariableNames',{'XX','YY','ZZ'})

T = 72×3 table
      XX         YY         ZZ   
    _______    _______    _______

    0.93769    0.92456    0.99817
    0.86066    0.90427    0.99263
    0.84509    0.85719    0.98951
     0.8738    0.89004    0.99273
    0.55918    0.68187    0.55212
     0.7432    0.56476    0.70838
    0.67489    0.51727    0.64014
      0.764    0.72324     0.6591
    0.81925    0.90581    0.90397
    0.71105    0.74472    0.83758
    0.71823    0.71584    0.84393
    0.90053    0.85199    0.96139
    0.90284    0.82024    0.98977
    0.78653    0.76117    0.97559
    0.75978    0.70471    0.97234
    0.78341    0.70018    0.97691

T = splitvars(struct2table(ICCs),'Metrics','NewVariableNames',{'XX','YY','ZZ'})

T = 72×3 table
      XX         YY         ZZ   
    _______    _______    _______

    0.93769    0.92456    0.99817
    0.86066    0.90427    0.99263
    0.84509    0.85719    0.98951
     0.8738    0.89004    0.99273
    0.55918    0.68187    0.55212
     0.7432    0.56476    0.70838
    0.67489    0.51727    0.64014
      0.764    0.72324     0.6591
    0.81925    0.90581    0.90397
    0.71105    0.74472    0.83758
    0.71823    0.71584    0.84393
    0.90053    0.85199    0.96139
    0.90284    0.82024    0.98977
    0.78653    0.76117    0.97559
    0.75978    0.70471    0.97234
    0.78341    0.70018    0.97691

[Stephen23]

@Tomaszzz: did my answer resolve your question?

Answer 2

[KSSV]

load('data.mat')
name = reshape(struct2array(ICCs),3,[])' ;
T = array2table(name);

Or you can use

load('data.mat')
T = struct2table(ICCs);

[Stephen23]

Avoid this code, it is not robust. Consider what would happen if the structure contains other fields.

One of the main benefits of using structures is that each field can be processed independently. This approach using STRUCT2ARRAY discards that benefit, making the code more fragile to future data changes.

EDIT: after KSSV editted and added a second approach: the second code suggestion does not work because the OP requested three columns in the table, and that code only returns one.

Sigh.

Answer 3

[Mathieu NOE]

hello

this would be my suggestion

 load data.mat
 
 % convert structure to numerical array
 a = struct2cell(ICCs);
 a = squeeze(a);
 a = vertcat(a{:});
 
 % create table
 t = array2table(a,'VariableNames',{'name 1' 'name 2' 'name 3'});

[Stephen23]

One of the main benefits of using structures is that each field can be processed independently. This approach using STRUCT2CELL discards that benefit, making the code more fragile to future data changes.

SQUEEZE (which is an abomination anyway) is not required if the variable is used for a comma-separated list.