creating date for data sorting in time series data

Hi everyone,

I have a time series of sea level with unevenly time interval as you can see in the following:

10-03-2014 17:05 -26.86
10-03-2014 17:07 -26.85
10-03-2014 17:09 -26.85
10-03-2014 17:11 -26.86
10-03-2014 17:13 -26.85
10-03-2014 17:15 -26.87
10-03-2014 17:17 -26.86
10-03-2014 17:19 -26.86
10-03-2014 17:21 -26.86
10-03-2014 17:23 -26.86
10-03-2014 17:25 -26.86
10-03-2014 17:27 -26.85
10-03-2014 17:29 -26.86
10-03-2014 17:31 -26.84
10-03-2014 17:33 -26.85
10-03-2014 17:35 -26.86
10-03-2014 17:41 -26.86
10-03-2014 17:45 -26.87
10-03-2014 17:47 -26.88
10-03-2014 17:49 -26.87
10-03-2014 17:55 -26.87
10-03-2014 17:57 -26.84
10-03-2014 17:59 -26.86
10-03-2014 18:01 -26.83
10-03-2014 18:03 -26.83
10-03-2014 18:05 -26.83
10-03-2014 18:07 -26.84
10-03-2014 18:11 -26.84
10-03-2014 18:13 -26.86

now I want to have data in every 10 minutes. I think that it's good that I generate the data every 1 minutes and after that I just keep the data every 10 minutes. so I want to generate the data every one minutes and if we do not have data in that time the code write nan instead.

Kindly someone helps how to do this. Many many thanks in advance.

Answer 1

[Walter Roberson]

If you have R2016b or later, it might be easiest to create a timetable() and use retime()

[Mina Masoud]

unfortunately, I have 2014 version of Matlab

[Walter Roberson]

Presuming that the time information is stored in a cell array of character vectors named "times", then

times = {
    '10-03-2014 17:05'
    '10-03-2014 17:07'
    '10-03-2014 17:09'
    '10-03-2014 17:11'
    '10-03-2014 17:13'
    '10-03-2014 17:15'
    '10-03-2014 17:17'
    '10-03-2014 17:19'
    '10-03-2014 17:21'
    '10-03-2014 17:23'
    '10-03-2014 17:25'
    '10-03-2014 17:27'
    '10-03-2014 17:29'
    '10-03-2014 17:31'
    '10-03-2014 17:33'
    '10-03-2014 17:35'
    '10-03-2014 17:41'
    '10-03-2014 17:45'
    '10-03-2014 17:47'
    '10-03-2014 17:49'
    '10-03-2014 17:55'
    '10-03-2014 17:57'
    '10-03-2014 17:59'
    '10-03-2014 18:01'
    '10-03-2014 18:03'
    '10-03-2014 18:05'
    '10-03-2014 18:07'
    '10-03-2014 18:11'
    '10-03-2014 18:13'};
sea_levels = [
    -26.86
    -26.85
    -26.85
    -26.86
    -26.85
    -26.87
    -26.86
    -26.86
    -26.86
    -26.86
    -26.86
    -26.85
    -26.86
    -26.84
    -26.85
    -26.86
    -26.86
    -26.87
    -26.88
    -26.87
    -26.87
    -26.84
    -26.86
    -26.83
    -26.83
    -26.83
    -26.84
    -26.84
    -26.86];

    dn = datenum(times);
    min_time = min(dn);

    min_time_dv = datevec(min_time);
    min_time_dv(5) = floor(min_time_dv(5) / 10) * 10;  %move to beginning of 10 minute slot
    first_slot_dn = datenum(min_time_dv);

    max_time = max(dn);
    max_time_dv = datevec(max_time);
    max_time_dv(5) = floor(max_time_dv(5) / 10) * 10 + 10;  %move to end of 10 minute slot
    last_slot_dn = datenum(max_time_dv);

    ten_mins_as_days = 1 / (24 * 60 / 10);
    slot_dns = first_slot_dn : ten_mins_as_days : last_slot_dn;
    slot_ds = datestr(slot_dns);

    [~, slot_idx] = histc(dn, slot_dns);
    mean_sea_levels = accumarray(slot_idx, sea_levels(:), [length(slot_dns)-1,1], @mean, nan);

    output = [cellstr(slot_ds(1:end-1,:)), num2cell(mean_sea_levels)];

[Mina Masoud]

Thank you so much. I really appreciate that.

[Walter Roberson]

Note: due to round-off error in constructing the slot boundaries, slot_dns, , it is possible that times very close to a ten-minute boundary might get put into the wrong side of the boundary. If that was important, it could be compensated for.