How to change the folder path in FOPEN for creating .dat file

i got an example code, but i dont understand how the path works.

1. fid= fopen(['.\result\norm\' algo '\' ActFunc_Hid '_' ActFunc_Out '\NOUT' num2str(nout) '.dat'], 'a'); %folder path
2. fprintf(fid, '\n \n');
3. fprintf(fid, '%s \n',algo);
4. fprintf(fid, '%s %s \n',ActFunc_Hid,ActFunc_Out);
5. fprintf(fid, '%s %s %s %s %s %s \n','nhid','repeat','TrainAcc','ValAcc','TestAcc','time');

assume algo: algorithm, ActFunc_Hid : hidden layer active function, ActFunc_Out : output layer active function, NOUT : neuron output, nhid is hidden neuron, etc

For 1, why does it start with a '.\' and what does those '_' and '\' in between mean?

Question: How to change the folder path?

Answer 1

[Walter Roberson]

Replace the first line with:

folder = fullfile('result', 'norm', algo, [ActFunc_Hid '_' ActFunct_Out]);
if ~exist(folder, 'dir')
    mkdir(folder);
end
filename = fullfile(folder, sprintf('NOUT%g.dat',nout));
fid = fopen(filename, 'a');

why does it start with a '.\'

In MS Windows, that means that you want to refer to something within the current folder.

and what does those '_' and '\' in between mean

The '_' is not special: it is just a literal underscore, used to separate the Hid value from the Out value so that they are easy to read and do not run together.

In MS Windows, the '\' is a directory seperator character, marking the end of a previous subdirectory name. After a '\' there can be another subdirectory name, but the last part can instead be a file name.

MacOS and Linux use '/' instead of '\', and it turns out that MS Windows can use '/' as well.

fullfile() knows how to use '\' or '/' as appropriate for the operating system being executed on.

[Halsey]

Thank you for your answer, i have understood now! However, i have no clue when it comes to create a .dat file, that related to num2str

fid= fopen(['.\result\norm\' algo '\' ActFunc_Hid '_' ActFunc_Out '\NOUT' num2str(nout) '\RESULT_myANN_pcvar_' num2str(var) '.dat'], 'a'); %folder path

[Walter Roberson]

folder = fullfile('result', 'norm', algo, [ActFunc_Hid '_' ActFunct_Out], sprintf('NOUT%d',nout)  );
if ~exist(folder, 'dir')
    mkdir(folder);
end
filename = fullfile(folder, sprintf('RESULT_myANN_pcvar_%d.dat') );
fid = fopen(filename, 'a');

[Walter Roberson]

Which is to say:

• in any case where you would have a directory separator, use fullfile() instead
• It is cleaner and more powerful to use sprintf('CONSTANT%format', value) rather than ['CONSTANT' num2str(value)]
• When you are writing output files with directory names that are computed, then it is quite common that your target directories will not exist, so mkdir() is often recommended

[Halsey]

That's clear, Thank you for your detailed explanations.