If you want the rolling average, use smoothdata, with 'movmean' as the method and (e.g.) hours(6) as the window. If you want non-overlapping hourly/6-hourly averages, use retime. There will be two different ways to use retime, as hourly averaging is built-in:
retime(tt,'hourly','mean')
whereas you'll want to specify the time vector or timestep for the 6-hour averages. If you just specify the timestep, the 6-hour bins start from the start of your data:
retime(tt,'regular','mean','TimeStep',hours(6))
or
retime(tt,datetime(2010,5,31,12,0,0):hours(6):<whatever end time>,'mean')
