Subdividing a matrix and pairing specific columns in the subdivisions

Hi experts,

I have a matrix of 1001 columns and 6000 rows. The first column is frequency, and the rest are signal intensities of a 1000 different samples. I want to create individual xy columns for the 1000 samples (so 1000 individual xy files), with the first column as x in each and the second column as each subsequent y value.

So D=[x y1 y2 y3 .....y1000], where each of these entries is a column of 6000 datapoints and I would like xy, xy1, xy2, xy3....xy1000 where xy=6000 x 2.

How would I go about this please?

Thank you in advance!

Best wishes,

Michel

Answer 1

[Image Analyst]

Try this:

% Create sample data
data = rand(6000, 4);
% Sort first columns
data(:, 1) = sort(data(:, 1), 'ascend');
% Now we have our data and can begin.
[rows, columns] = size(data)
freq = data(:, 1); % Frequency is column 1.
for col = 2 : columns
	fileName = sprintf('Column %3d.txt', col);
	fprintf('Writing column %d to "%s".\n', col, fileName);
    % Create N-by-2 matrix of frequencies in column 1, then signal in column 2.
	xy = [freq, data(:, col)];
	% Write out data to the text file.
	writematrix(xy, fileName);
end

[Michel Nieuwoudt]

Thank you Image Analyst - this is just perfect!

I would like to ask another question in extesnion of this please:

If I already have a .csv file with the names of each column, in the same order as the order in the matrix; would it be possible to save each column as that filename, instead of column2.txt, column3.txt,..etc?

I suspect so, but how would I do this please?

Best wishes,

Michel

[Image Analyst]

Yes. Let's say that the names you already have are stored in a cell array variable called columnNames. Then just use that for the filename:

% Now we have our data and can begin.
[rows, columns] = size(data)
freq = data(:, 1); % Frequency is column 1.
for col = 2 : columns
	fileName = sprintf('%s.csv', columnNames);
	fprintf('Writing column %d to "%s".\n', col, fileName);
    % Create N-by-2 matrix of frequencies in column 1, then signal in column 2.
	xy = [freq, data(:, col)];
	% Write out data to the text file.
	writematrix(xy, fileName);
end

[Michel Nieuwoudt]

Thank you! Your help and input is very much appreciated.

Best wishes,

Michel

Answer 2

[the cyclist]

You should strongly reconsider whether you need dynamically named variables, because it is usually a terrible idea. Instead, consider making a 3-dimensional array. Then, instead of the variable xy7, you can just access xy(:,:,7). Your code will be more robust.

% Pretend input
D = rand(6000,1001);

% Initialize 3-dimensional array
xy = zeros(6000,2,1000);

% Fill first "slice" in 2nd dimension with copies x
xy(:,1,:) = repmat(D(:,1),[1 1 1000]);

% Fill second slice with y
xy(:,2,:) = reshape(D(:,2:end),6000,1,1000);

[Image Analyst]

Do you think he really meant that, because like you said, that would be a terrible idea?

I thought he wanted to just save them (each xy 6000-by-2 matrix) to 1000 files or somewhere rather than in 1000 separate, individually named variables.

[Michel Nieuwoudt]

This is true, I do need to have separate files that I can plot individually in a spectroscopic software that won't plot matrices, so that I can perform band resolutions, etc. where needed.

Thank you to the cyclist for the suggestion though. I very much appreciate your answers!

Best wishes,

Michel