Anthony DeRobertis wrote:
> On Wed, Feb 08, 2006 at 05:13:57PM -0800, John Pettitt wrote:
>   
>> All    N=169          Delay     Offset   Jitter                  
>> Median                81.259     0.245     1.378
>>     
>
> This seems to say that the median of the entire data set is closer to
> reality than the median of any of its subparts... That fails the sanity
> check.
>
> I think this is from the stratum 1's having a negative median. What
> happens if you take the absolute value of each offset first?
> _______________________________________________
>   
The point was to find out what the collective view of the time was and
since some art positive and some are negative the median can be
negative.  If I took the absolute I can see that it would tell me how
far a typical pool server is from my S1 - maybe I'll do that in the next
generation of stats.  I'm still playing with what to graph and how to
represent it (If there are any stat;s junkies out there who want to
offer suggestions on meaningful things to graph please let me know).

John
_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to