Better way to keep dimensions using reshape from extracted struct array?

HEllo

I am wondering what is the best way to extract the data from an structured array in matlab and keeping is dimensionality if I use reshape while processing it.

Let's say I have the following struct array

load('data.mat')
myData.data.b

ans = 1x2

20.0183 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ans = 1x2

20.0183 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ans = 1x2

20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ans = 1x2

20.0183 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

ans = 1x2

20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

In order to extract all b values I would do

[myData.data(:).b]

ans = 1x10

20.0183 19.8833 20.0183 19.8833 20.0182 19.8833 20.0183 19.8833 20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

But this rises the issue that the dimensionality of b is lost, since now it is treated as an vector. If I do reshape to try to recover that dimensionality I will get

bb=reshape([myData.data(:).b]',[],2)

bb = 5x2

20.0183 19.8833 19.8833 20.0183 20.0183 19.8833 19.8833 20.0182 20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

which is still not the original data, because all even row are swapped compared to original. So the only thing I came up was to undo that swapping on even rows via

bb(2:2:end,:) = fliplr(bb(2:2:end,:))

bb = 5x2

20.0183 19.8833 20.0183 19.8833 20.0183 19.8833 20.0182 19.8833 20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Hence I wonder whether there exists a better way to extract b from the struct array without having to do the reshape + even elemen flip.

thanks in advance,

Bes regards

[VBBV]

@Javier Cuadros, you can try using a loop to extract data from nested struct array

load('data.mat')
for k = 1:numel(myData.data(:))
    mD(k,:) = myData.data(k).b;
end
mD

mD = 5x2

20.0183 19.8833 20.0183 19.8833 20.0182 19.8833 20.0183 19.8833 20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[Javier Cuadros]

Hi VBBV thanks for your reply, I was looking for something not using forloops since the data b can be bulky in the end

[Stephen23]

Why not just use VERTCAT as Voss shows?

https://www.mathworks.com/matlabcentral/discussions/tips/847976-tutorial-comma-separated-lists-and-how-to-use-them

[VBBV]

@Javier Cuadros,, Do you mean data size of field b is not constant ? if it has homogenous size, then as @Voss . @Stephen23 mentioned its better to use vertcat function

[Javier Cuadros]

@VBBV indeed, it is not constant it can grow in colums or rows

[Javier Cuadros]

thanks guys, Voss solution works! I didn't know that I could use it in that way!

[Stephen23]

"I didn't know that I could use it in that way!"

You can use comma-separated lists with any function.

Answer 1

[Voss]

load('data.mat')
bb = vertcat(myData.data.b)

bb = 5x2

20.0183 19.8833 20.0183 19.8833 20.0182 19.8833 20.0183 19.8833 20.0182 19.8833

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[Javier Cuadros]

Thanks @Voss this works !!

[Voss]

You're welcome!

Answer 2

[arushi]

Hi Javier,

To extract data from a structured array in MATLAB while preserving its original dimensionality, you can use a combination of array manipulation functions that maintain the structure of your data. In your case, since each element of myData.data(:).b is a 1x2 vector, you can use reshape and cell2mat in a way that respects the original organization of the data.

Here’s a more efficient approach to extract b values from the struct array without manually flipping rows:

1. Extract Data into a Cell Array: Use arrayfun to extract the data into a cell array, preserving the structure.
2. Convert to a Matrix: Use cell2mat to convert the cell array into a matrix.

       bMatrix = cell2mat(bCellArray');
       bOriginal = reshape(bMatrix, [], 2);

Hope this helps.

[Javier Cuadros]

HI Arushi, thanks for replying!

I don't understand fully step 1 because the extraction is the issue. Do you mean that my arrayfun needs to iterate over the struct array and convert it into a cell? Otherwise I always get only one element...

load('data.mat')
b1=arrayfun(@(x) x.data(:).b,myData,'UniformOutput',false)

b1 = 1x1 cell array
    {[20.0183 19.8833]}