.mat file size keeping it small on compiling

Hi,

I'm compiling some .mat files and saving it, but the overall saved data is bigger than the sumer of the files being used.

Is this normal? Is there a way to keep it down?

The compiled file is 470MB, and the sum of the files being used is about 125MB

form=1; %make sure to also change the formulation number under filename
scenarios = 5;
Nscen = numel(scenarios);
IDsets = 1:500:3001;
NID = numel(IDsets);

filedir = 'Outputs';
indvoutputagg = cell(NID,1); % preallocate
indvoutputcomp={};
for i=1:Nscen
for k = 1:NID
    filename = sprintf('Form1_Scenario%d_ID%ds_indvoutputagg.mat',scenarios(i),IDsets(k));
    indvoutputagg{k,:} = importdata(fullfile(filedir,filename)); % allocate using indexing
    % Make it into one big table
    indvoutputcomp(1+size(indvoutputcomp,1):size(indvoutputcomp,1)+size(indvoutputagg{k,1},1),:)=indvoutputagg{k};
end
save(['Outputs/Form' num2str(form) '_Scenario' num2str(scenarios(i)) '_indvoutputagg.mat'],'indvoutputcomp')
end

[Walter Roberson]

Please check whether the output .mat file is a -v7 file or a -v7.3 file. Your code is not specifically asking for -v7.3 but you might potentially have set your preferences to use -v7.3 by default.

[In-chan Kim]

I set the preference to v7.3 because it was saying it's too large to save, unless it was v7.3

But even then, is it normal for a combined 125MB to become so large? Is there something in my coding that could be adjusted?

[Walter Roberson]

Yes, it is common for -v7.3 files to be large. Cell arrays in particular have more overhead than you would expect.

[In-chan Kim]

Is there something I can do to make it smaller?

[Walter Roberson]

I recommend against using importdata() -- too much overhead for the cases where the output is well defined, and too uncertain about what it will return for the other cases. You are reading in a .mat file: just load() it .

The .mat files you are reading in: do they all have exactly the same variable names? Would the variable names also happen to be in exactly the same order in each file? If they do not have exactly the same variable names, then are the variable names self-identifying as to which file they came from? For example if all of the 501 variables mention 501 at a particular location in their name, then if we threw all the loaded variables into the same structure, we could figure out afterwards which file they came from.