Hi Cedric,
If you already have your dates in a cell array string format, you should not need to convert to datetime in order to create a timeseries. For example, see the code below
%%Generate 100 random dates with a specified string format
datenumbers = rand(100,1)*now;
datenumbers = sort(datenumbers); % put them in increasing order
dateStrings = datestr(datenumbers, 'dd-mm-yyyy HH:MM:SS');
%%At this point "dateStrings" should be just like the data you have
Ts = timeseries(data, dateStrings);
Currently you cannot create a timeseries with "datetimes", but using a date string should work. If this is still slow for you, you may want to post either your data and full code or the results of running the profiler on the code (hit the "Run and Time" button).
Hope that helps,
-Adam
