Create a table from a list of variables

I'm loading a file that contains several Nx1 arrays of data, each one being a separate variable, plus various metadata

I want to assemble all the data into a table, how can I do that?

this is what I'm doing at the moment, it works, it's fast for what I'm doing, but I feel guilty because my grandma once told me never never to use eval

vars = whos;
vars = vars(strcmp({vars(:).class},'single'));
T = table;
for ii = 1:length(vars)
    eval(['T.' vars(ii).name '=' vars(ii).name ';'])
end

[Peter Perkins]

Your grandma is smart. You can avoid eval on the LHS of that assignment

T.(vars(ii).name) = ...

but you do need an eval for the RHS.

If you knew the workspace var names, and they were always the same

t = table(myvar1,myvar2,...)

would capture the workspace names, but presumably you are not in that boat. So Rik's advice is good.

Answer 1

[Rik]

If you are loading a file you should use this syntax:

S = load(filename);

That way, all variables in the mat file are stored as the fieldnames of S. This means it is always clear where variables came from. If you need the variable names for some reason, you can use the fieldnames function. struct2table may already do what you need.

[Paolo Mazzoleni]

thanks

I've tried and struct2table would work well if my struct contained only arrays of data, but in my case it also contains other fields that are metadata, all stored as strings

fieldnames just returns the name of the fields, how can I get the type of each field so that I can keep only those that are single arrays?

[Paolo Mazzoleni]

basically I have something like this

and I want to create this

but I actually have many columns and their name is different in different files

[Rik]

Just like you would otherwise: with dynicamic indexing.

example = struct(...
    'a',rand(10,1),...
    'b',single(rand(10,1)),...
    'c','some description',...
    'd',single(rand(10,1)));
f_names = fieldnames(example);
slim = example;
for n=1:numel(f_names)
    if ~isa(slim.(f_names{n}),'single')
        slim = rmfield(slim,f_names{n});
    end
end
disp(slim)

    b: [10×1 single]
    d: [10×1 single]

[Paolo Mazzoleni]

it works, thanks