Talk:Erlang distribution

From Wikipedia, the free encyclopedia

This article is within the scope of WikiProject Statistics, which collaborates to improve Wikipedia's coverage of statistics. If you would like to participate, please visit the project page.

The discussions on this page have been reogranized. I have split previous comments up and duplicated the signature lines when spliting comments and added braces for clarity. Acuster 06:18, 19 August 2005 (UTC)

Contents

[edit] Importance of the topic

The redirect from the Erlang-B page, via the Erlang-C reference link, is completely incorrect. As already noted several times, this page presents the Erlang_k or restricted Gamma distribution. The B and C functions are usually treated separately from the the k-distribution because they arise in an entirely different context in queueing theory. The correct content for Erlang-C exists at http://de.wikipedia.org/wiki/Erlang_C. Unfortunately, it's in German but the math is correct. —Preceding unsigned comment added by RedRooz (talk • contribs) 22:15, 2 February 2008 (UTC)

We do not need a long article here since the Erlang distribution is basically the Gamma distribution. vignaux 02:06, 2004 Sep 20 (UTC)

I agree that this article could be merged into gamma distribution (unless it can be significantly expanded to include a detailed discussion of the phone traffic scenario). --MarkSweep 17:17, 7 Nov 2004 (UTC)
It is true that they are all actually probability distributions but the Erlang Distribution is a continuous distribution, a special case of the gamma but with some special characeristics - it can be modelled as a sum of exponentials, for example. I think that it unique enough that it should have its own page as well as being referred to in the gamma distribution page as a special case. --vignaux 07:44, 2004 Nov 9 (UTC)
Ok, then the present article should be kept and revised to reflect the situation you describe. --MarkSweep 00:34, 10 Nov 2004 (UTC)
The Erlang is widely used and named as its own distribution. It deserves a page of its own with explicit reference to the Gamma. Algebraically, it is often possible to generate Erlangs in computer programs where it would much harder to generate Gammas since the factorial is widely defined but the Gamma functions are less frequently available. Acuster 06:18, 19 August 2005 (UTC)
I've found this page to be useful. More useful than the page on the Gamma distribution. One thing I like about this page is that it is explained, in simple terms, where the Erlang distribution comes from. It is relatively easy to understand, since, in part 'k' is an integer (event number). But the meaning of the Gamma distribution, for which the corresponding parameter is real, is a mystery to me. I say keep this page, but fix the Gamma distribution page, which is lacking any kind of motivational explanation.

[edit] Contents

[edit] Plots

I wanted to put plots in this article, but the Gamma distribution plots would work just as well, if only the two articles agreed on whether to use θ or λ as the parameter. It would also facilitate understanding the relationship between the two distributions. Is there any reason not to choose one and have both articles use the same parameter? PAR 09:48, 19 August 2005 (UTC)

  • There may be a reason. If I remember correctly when you do the math to derive the erlang from the exponential, you get one parameter. Then, the Gamma generally uses the other. I'll look into it. Acuster 05:29, 20 August 2005 (UTC)
I added plots from the Gamma distribution article, but the CDF does not appear in my browser. Does it look ok to everyone else?
Looks great to me, thanks. Acuster 21:53, 21 August 2005 (UTC)

[edit] The Poisson process

You should better explain Poisson process. vignaux 02:06, 2004 Sep 20 (UTC)

[edit] Erlang-B and Erlang-C

The material on the Erlang Formulae should be moved to another page-- it is not the Erlang Distributiion. vignaux 02:06, 2004 Sep 20 (UTC)

{However, w}{W}hat exactly do you mean by Erlang formulae vs. Erlang distribution? Would you care to elaborate? --MarkSweep 17:17, 7 Nov 2004 (UTC)

The Erlang-B and Erlang-C formulae are queue theory results giving the discrete probability distributions of the numbers in steady-state queues with certain characteristics (blocked -customers-cleared, for example). They are not usually referred to as Erlang Distributions. --vignaux 07:44, 2004 Nov 9 (UTC)
Ok, then the present article should be kept and revised to reflect the situation you describe. --MarkSweep 00:34, 10 Nov 2004 (UTC)
{t}The Erlang-B and Erlang-C formulae are mentioned in the Erlang Unit page. I will modify this page when I have time. --vignaux 04:01, 2004 Nov 12 (UTC)

[edit] Comments

[edit] More specific definition of cumulative would be welcome

I assume the posted formula for the cdf is over a range from 0 to x. Is this correct? I've found this integral difficult to perform, hence my question.

[edit] Please define constants in equation

Please define the constants in the equation. I would but I don't know what they are!

Huh? The parameter n is explained in the preceding text, x is defined by function abstraction, and e is Euler's number. Am I missing something here? --MarkSweep 03:47, 11 Jan 2005 (UTC)
sorry, I just couldn't find them.Pdbailey 23:04, 12 Jan 2005 (UTC)