How can I aggregate rows by a specified start time and time step in a time table?

I have 15-minute schedules and I want to aggregate time steps with given start time using mean, sum, max, and so on. I've already tried to use the retime function, but it only makes the sum to the next full hour and I need the sum of the next hour. For instance the sum of Load for the next hour starting at 14:30 and including the time steps 14:45, 15:00, 15:15 ...

TStep = hours(1);
AggregateSchedule = retime(Schedule,'regular','sum','TimeStep',TStep);

         DateTime              ID        Load    RL    Price
    ___________________    __________    ____    __    _____

    14.06.2018 14:30:00       7617       6148    0      250 
    14.06.2018 14:45:00       7617       6253    0      250 
    14.06.2018 15:00:00       7617       6358    0      250 
    14.06.2018 15:15:00       7617       6463    0      250 
    14.06.2018 15:30:00       7617       6568    0      250 
    14.06.2018 15:45:00       7617       6673    0      250 
    14.06.2018 16:00:00       7617       6778    0      250 
    14.06.2018 16:15:00       7617       6883    0      250

Many thanks in advance!

Answer 1

[Guillaume]

My suggestion would be to subtract 30 minutes from your datetime column, retime it as you've done then add the 30 minutes back:

Schedule.Properties.RowTimes = Schedule.Properties.RowTimes - minutes(30);
AggregateSchedule = retime(Schedule, 'hourly', 'sum');
AggregateSchedule.Properties.RowTimes = AggregateSchedule.Properties.RowTimes + minutes(30);

[s.h.m_89]

Many thanks for your workaround Guillaume!

I also found another solution, described in the following link:

https://de.mathworks.com/matlabcentral/answers/352778-would-be-possible-to-calculate-half-hour-average-by-using-retime

[Peter Perkins]

As I suspect that link describes, retime will happily aggregate on the half-hours, it's just a little bit more work than using 'hourly'. Create an explicit target time vector as something like

target = datetime(2018,6,14,14,30,0):minutes(30):datetime(...)

Also, beginning in R2018a, there's a new syntax to make that easier:

tt = retime(tt,'regular',method,'TimeStep',minutes(30))