Grouping based upon differences in time, with some small scatter

Hello!

We have an application to analyze digital signals from several events on separate instruments. There is a time stamp in each record. There can be a small offset in the time recorded (<30 seconds) for a single event on separate instruments. For example:

{[27-Aug-2020 12:30:13]}

{[27-Aug-2020 12:39:23]}

{[27-Aug-2020 12:47:46]}

{[27-Aug-2020 12:30:15]}

{[27-Aug-2020 12:39:25]}

{[27-Aug-2020 12:47:48]}

{[27-Aug-2020 12:30:18]}

{[27-Aug-2020 12:39:28]}

So there are three events (~12:30, ~12:39, ~12:47), the first two recorded on three instruments, the last event on only two. Since the delay may be up to 30 seconds, ignoring the seconds isn't an option.

In practice there may be tens of events and a hundred instruments, so grouping this is a bit of a pain. I can use datenum() and then seconds() to get whether the difference is less than 30 seconds, but kind of hung up on what to do next.

Any suggestions?

Thanks!

Doug Anderson

[Douglas Anderson]

A thought...

findgroups() looks like it might be helpful, but not sure how to make a group +- 30 seconds

[Douglas Anderson]

Or maybe use of uniquetol()??

Answer 1

[Duncan Po]

One way to do this is to use isbetween in a for loop, and then unique(, 'rows') to find the groups:

t = datetime({'27-Aug-2020 12:30:13',

'27-Aug-2020 12:39:23',

'27-Aug-2020 12:47:46',

'27-Aug-2020 12:30:15',

'27-Aug-2020 12:39:25',

'27-Aug-2020 12:47:48',

'27-Aug-2020 12:30:18',

'27-Aug-2020 12:39:28'}); % put the data in a datetime

tmax = t + seconds(30); % determine the 30s offsets from each element

tmin = t - seconds(30);

for i = 1:length(t) % loop through each element

y(i,:) = isbetween(t, tmin(i), tmax(i));

end

yy = unique(y, 'rows') % each row is a logical vector indicating which element belongs to the same group

[Douglas Anderson]

Thank you Duncan!

Works great.