Brilliant, thanks Derek.

Scott

On 05/05/2012, at 8:36 AM, Derek Lamb wrote:

> Hi Scott,
> 
> Assuming your $t is always increasing (you can qsort() it if it's not), you 
> can assemble a vector of delta-t's, and if you're only interested in integer 
> offsets, use the hist function to generate a histogram of sorts:
> 
> $dt = $t(1:-1) - $t(0:-2);
> ($interval,$number) = hist($dt,$dt->min-0.5,$dt->max+0.5,1);
> p $interval,$number;
> p $number->where($interval==50);
> 
> Now you have one piddle that contains a list of the integer offsets, and 
> another piddle that tells you the frequency.  If you have non-integer times, 
> then that will require some more thought as to how close two differences can 
> be and be considered the same. 
> 
> I've also discovered that Maggie's PDL::Stats (available from CPAN) might 
> have things that help with the time-series analysis you seem to be doing (see 
> the PDL::Stats::TS module).  For example, I recently had to take a moving 
> average of a time series of data, then compute the point-to-point differences 
> of that moving average, and it was as simple as:
> 
> use PDL::Stats;
> $data->filter_ma(3)->diff;
> 
> best,
> Derek
> 
> On May 4, 2012, at 4:17 PM, Scott Penrose wrote:
> 
>> Good morning
>> 
>> I have a piddle with time stems, e.g.
>> 
>> print $t;
>> [ 100, 150, 200, 249, 300 ];
>> 
>> The data is really long and obviously the example above is made up.
>> 
>> Anyway I want to calculate the number between and calculate, so from the 
>> data above I would have
>> 
>> 50 = 2
>> 49 = 1
>> 51 = 1
>> 
>> The main reason I am doing this is to make sure that my measurements are on 
>> the window (in this case 50 ms) and how many are not.
>> 
>> Currently I just did it by hand, keeping the last value and iterating 
>> through all the data in a loop. But it didn't feel very PDL way. What is the 
>> better way?
>> 
>> Ta
>> 
>> Scott
>> 
>> 
>> _______________________________________________
>> Perldl mailing list
>> [email protected]
>> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
>> 
> 


_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to