How to select specific data from a structure

Hi, I have the following structure: data.variation_#.ts

- data has 23 variation fields (i.e. variation_1, variation_2 etc.), and I'm interested in field 3 to 23

- The ts field is a 30x10800 double

How can I read throug the variation fields of interest(3 to 23) and for all of them get some of the rows from ts (for example rows 2 3 4, all columns)

Thanks

[dpb]

"... following structure: data.variation#.ts"_

Bad idea here is your basic problem. Instead of sequentially-named fields of 2D array, use 3D array where each variation is a plane. Then you simply refer to the planes of interest.

Answer 1

[Guillaume]

This is why I think it's not a good idea to recommend using dynamic field names instead of eval, it leads to the same problems.

If you have numbered variables or field names, you have a poor data structure and it indeed becomes difficult to operate the same process on all of them at once. The proper solution is to get rid of all these numbered variables / field names and store their content into a cell array (if the size of the content varies from variable to variable) or a matrix (if not). It is then trivial to index them.

See Jos' answer for a good structure. However, personally I would avoid the multilevel structure, it's a pain to work with (See my comment to Jos' answer).

Assuming that the variation fields are the only field of data, you can convert your existing structure into Jos' suggestion with:

newdata.variation = cell2mat(struct2cell(data))
%or you could just do
variation = cell2mat(struct2cell(data))
%and not bother with multilevel structures

To then get rows [3 4 8] of variations 3 to 23 as a 3d array:

selectedts = cat(3, variation(3:23).ts);
selectedrows = selectedts([3 4 8], :, :);

[Isma_gp]

Hei, the variation fields are not the only fields of data. Is there any way to convert the data then?

Thanks

[Guillaume]

Probably, the safest way to do this:

wantedfields = sprintfc('variation_%d', 3:23);
[~, fieldrows] = ismember(wantedfields, fieldnames(data));
dataascell = struct2cell(data);
variations = cell2mat(dataascell(fieldrows));

selectedts = cat(3, variation(3:23).ts);
selectedrows = selectedts([3 4 8], :, :);

Answer 2

[Jos (10584)]

Store your data like this

data.variation(1).values = ..
data.variation(2).values = ..
data.variation(3).values = ..

which makes it trivial to select the fields you need

data.variation([3 4 8])

It is the content of a variable (or field) that is supposed to be flexible, not its name!

[Guillaume]

Totally agree with "It is the content of a variable (or field) that is supposed to be flexible, not its name!"

The proposed solution however does not work as is, since

data.variation([3 4 8]).values

will return a comma-separated list. Some extra machinery is required, either:

cat(x, data.variation([3 4 8]).values); %where x is ndims(values)+1

if all the values have the same size or

{data.variation([3 4 8].values}

if not.

[Jos (10584)]

Nice follow-up, Guillaume.

[Shahram]

Thanks "data.variation([3 4 8])" worked wel