Tricky correction of faulty time list

Hi. I have a Nx1 cell array containing different time values as strings, mixed wildly. It looks something like this:

12:00
12:05
13:45
09:15
08:25
*08:32*
15:20
*15:26*

but a lot longer... As you can see, the minute values are almost all multiples of 00:05. But some random ones are shifted by one or two minutes.

Is there a way I can run through the array and correct the shifted values? So in my example I would want to change '15:26' to '15:25' and '08:32' to '08:30'.

If it is not possible to automate it, finding them would be also be okay - so that I could display their respective row numbers in a message box to be able to correct them manually.

Thanks in advance for any help!

Answer 1

[Azzi Abdelmalek]

v={'12:00';'12:05';'13:45';'09:15';'08:25';'08:32';'15:20';'15:26'}
c=cellfun(@(x) str2double(regexp(x,'(?<=\:).+','match')),v);
c1=round(c/5)*5
for k=1:numel(v)
  v{k}=datestr(datenum([v{k}(1:3) num2str(c1(k))],'HH:MM'),'HH:MM');
end
v

%Or, using Jan's approach

v = {'12:00';'12:05';'13:45';'09:15';'08:25';'08:32';'15:20';'15:26'};
v=regexprep(v,'(?<=\d+\:\d)[1-4]','0')
v=regexprep(v,'(?<=\d+\:\d)[6-9]','5')

[Marc Jakobi]

Thanks to both for the awesome help! I decided to go for your second answer. All of them seem to work, but your second one and Jan's second approach are the fastest options.

[Azzi Abdelmalek]

regexp is slow. Both Jan's answers are faster then mine

Answer 2

[Jan]

v = {'12:00';'12:05';'13:45';'09:15';'08:25';'08:32';'15:20';'15:26'};
for k = 1:numel(v)
  w = v{k};
  if w(5) < '5'
    w(5) = '0';
  else
    w(5) = '5';
  end
  v{k} = w;
end

[EDITED] Less readable:

for k = 1:numel(v)
  w    = v{k};
  w(5) = '0' + 5 * (w(5) >= '5');  % Implicit conversion to CHAR
  v{k} = w;
end

[Marc Jakobi]

Thanks to both for the awesome help! I decided to go for Azzi's second answer. All of them seem to work, but your second approach and Azzi's second one are the fastest options.

[Jan]

I've suggested this method for another reason that the run time: It is simple and easy to debug. If a reader needs 5 seconds less to understand it, it does not matter, if it runs in 0.005 or 0.002 seconds.