RobertNiles.com
About Robert
Reporting Help
Finding Data on the Internet
Questions for Candidates
Stats Lessons
Mean
Median
Percent
Per capita
Standard Deviation
Margin of Error
Data Analysis
Sample Sizes
Stats Tests
Help Board
Bookstore


Statistics for the Utterly Confused

This book offers a super-accessible approach to the much-misunderstood subject of statistics.
More information
-->

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

Meaningful varience from Cisco RTR agent (data networks)?
Message posted by Stuart (via 166.44.137.198) on November 3, 2001 at 3:13 PM (ET)

I have an application that can gather statistics on round trip times in a data network.

It gives:
Number of Samples
Sum of Samples
Sum of Samples^2
(also Min & Max)

The samples are not recorded.

From those I can derive a Varience if it was normal distribution, but in data networks, the lower bound will be reasonably fixed.

Call it half a bell curve extending right with maybe "small" left portion.

It will never be less than A, normally be A + a bit, and will stretch to A + Variance.

I'm sure this isn't too hard but too long since Stats courses :-/

Regards,
S.


READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Meaningful varience from Cisco RTR agent (data networks)?
Message posted by Tomi (via 154.32.143.41) on November 3, 2001 at 11:38 PM (ET)

The formula you have probably used for variance is true for all distributions, regardless of normality. I assume you have used (sum of samples^2)/n - ((sum of samples)/n)^2.

Presumably you want a model that will help you make predictions.

What you are describing could be a truncated normal distribution (tricky to calculate but possible). More likely it is a Poisson distribution, as these are common in data transmission problems.

To decide on the best model I would need to see the data. You could send me an email with .csv or .txt attachment to t.owens@hautlieu.sch.je

I teach Statistics and it would be interesting for my students to consider this if you don't mind them seeing your data.


Re: Meaningful varience from Cisco RTR agent (data networks)?
Message posted by Shang-Pin Chang (via 163.28.64.112) on December 9, 2001 at 1:34 PM (ET)

I just found a gamma distribution does a not bad job in modeling delay by using Cisco's SAA (aka RTR) statistics.

If you're also interested in its jitter statistics, Laplacian distribution looks good.

Shang-Pin
--
Cisco Systems, Inc.



Your $5 contribution helps cover part the $500 annual cost of keeping this site online.

Niles Online Sites:RobertNiles.comTheme Park InsiderViolinist.com

RobertNiles.com™, the site, content and services 咀opyright 1996-2002, Robert Niles.
All rights reserved. Questions? Comments? Read my Privacy Policy, or E-mail me!