How to select a file based on the filename?

Hi, I have several files and I want to just select one based on the maximum number which I could find on the file name. For example I have

    abcEFG_1all_0.0675_0.2035.mat
    abcEFG_1all_0.087534_0.40354.mat
    abcEFG_1all_0.167534_0.603123.mat

and I just want to select and work on this file

abcEFG_1all_0.167534_0.603123.mat

because the numbers are higher than the others. Any suggestion would be appreciated in advance.

Answer 1

[Walter Roberson]

http://www.mathworks.com/matlabcentral/fileexchange/47434-natural-order-filename-sort

[Rita]

Thanks Walter for your quick response unfortunately,

Site Temporarily Unavailable

Due to planned maintenance, our site is currently unavailable. Please come back later.

Thanks

[Walter Roberson]

Is the "0." part of the number?

Is only the first number of the two to be paid attention to? Or are both numbers in the file name to be paid attention to?

[Rita]

yes 0. is a number .Actually the first one would be much more important.Thanks Walter

[Walter Roberson]

dinfo = dir('abc*.mat');
filenames = {dinfo.name};
parts = regexp(filenames, '_', 'split');
part3 = cellfun(@(C) C{3}, parts, 'Uniform', 0);
val3 = str2double(part3);
[maxval3, maxidx3] = max(val3);
desired_file = filenames{maxidx3};

[Rita]

Thanks a lot.You are amazing!!!

[Rita]

Hi Walter, I stuck in this script I followed your answer but for this one I get the error.I really would appreciate if you could solve my problem. I have

    net1_0.2327.mat
    net1_0.3425.mat
    net2_0.8765.mat 
    net2_0.6754.mat ,......to net50_0.87654.mat

I would like to get these net1_0.3425.mat net2_0.8765.mat ,.... the max of each net1 to net net50(comparing net1 with net1 and net2 with net2 and so on) I used your script

dinfo = dir('net*.mat');        
    filenames = {dinfo.name};
    parts = regexp(filenames, '_', 'split');
part2 = cellfun(@(C) C{2}, parts, 'Uniform',0);
parts3 = regexp(part2, '.mat', 'split');% to remove mat extension
val2 = str2double(parts3)% I get NaN instead of numbers!

Thanks in advance for your help.

[Walter Roberson]

Answered in the new Question you opened about this.