Finding intervals of data from repeated time stamp/marks

I have three sets of data times. They contain all the time stamps from a set of data, randomly divided into the three matrices (times_1,times_2, times_3). The recording took time stamps every .001 seconds. Much of these three matrices contain repeating time points, where consecutive .001 second time stamps are together. For those repeating or consecutive time points in each matrix, I would like to convert them into an interval style data point.

For example, if I have [43.5196, 43.5206, 43.5216,43.5226], I would like that to be converted into 43.5196-43.5226 or [43.5196, 43.5226].

Please let me know if there is any way to clarify this anymore!

Thanks for any help!

[dpb]

Convert to table or timetable and use grouping variable on the time data (as datetime) to compute the statistics of interest with retime or splitapply.

BTW, creating sequentially-numbered variables is a counterproductive way to code in Matlab in general; use one of the other storage mechanisms such as cell array or variable-named fields in a structure or somesuch so can write generic code without having to repeat the same code multiple times for the different data.

[Systematically Neural]

I usually try to do this, but I had to index for different variables for an entire data set of time stamps

[Systematically Neural]

Isn't converting to a table very computationally heavy?

[dpb]

So what else would you suggest doing? You've got to do the grouping in some manner; may as wellst use the tools TMW has developed for you instead of reinventing the wheel.

How big are you talking of?

"but I had to index for different variables for an entire data set of time stamps"

I have no idea what you're trying to explain there, but the more "alike" stuff there is, the more useful to try to generalize instead of making new variables.

[Systematically Neural]

Does the TMW tools have a function to group consecutive data points into intervals?

I have data sets 1 to 3 hours long with 1000 sampling freq, so about 10 million or so. The data have been indexed into three different possible states (1,2,3) then the time intervals containing each value extracted, which is what I attached.

I agree about the storage suggestion and will implement those, but it doesn't really address the question I am needing assistance on.

[dpb]

Well, after reading the other posting and this one again, you're trying to do something entirely different than what I thought your intent was -- the use of "repeated time" in the Q? title is a real red herring sending the respondent off in totally wrong direction....there are no repeated times at all so grouping by timestamp isn't what are trying to do at all.

accumarray may be more amenable to your task; the other Q? answer to find locations with breaks works.

>> ix=find(diff(t2)>0.001+eps(t2(1)));  % find breaks more than rounding error
>> length(t2)/length(ix)                % what's average length
ans =
    6.9783
>> ix(1:10)                             % where are first several
ans =
        1366
        3349
        4099
        4104
        4109
        4113
        4118
        4123
        4128
        4133
>>

OK, the first 1367 values are consecutive; so what do you want to do with them and what's the result to be? And where's whatever data goes with the timestamps?

[jonas]

Apologies, I posted my answer in the wrong thread. The fact that you've submitted two almost identical questions is a bit confusing.