Statistical distributions parameters explanation.

[gasser_y]

I am working with a 5-workstation assembly line. Each workstation follows a different statistical distribution; Gamma, Normal, 3-parameter weibull, Largest extreme value and Smallest extreme value (which by the way I can't find in the statistical distributions in FlexSim).

The thing is, I only understand the parameters of the Normal distribution (Mean, Standard Deviation). But I have no clue of the rest of the parameters for the other distributions. Parameters are usually: Location, Scale, and Stream.

Can I get any help, please ? What is meant by each parameter, and what are the expected values to be entered ??

And, what can be the alternative for "Largest extreme value" and "Smallest extreme value" distributions ??

Thank you so much for your time and efforts.

Best Regards,
Gasser Yassen

[joerg_vogel_HsH · Accepted solution]

Some answers now; Stream is the parameter of the random number stream. Please look for yourself in the manual or the Google search of this site to get more information of the random number generator.

Location is a static offset of the distribution value you get. It adds the value to the distribution value.

Scale enhances or compresses the result of the normalized distribution.

The uniform distributions are set in the range of minimum to maximum extrem values. Please search in the command manual for the keyword uniform in the index pages.

Maybe you get a detailed answer later.

[sam_stubbsYXX86]

Unfortunately I'm not a statistician, so I'm not sure I can explain the ins and outs of each distribution curve and how they work. But you can always get a more detailed explanation of each Flex Script command that you call. Any Flex Script command that you call (including any statistical distribution commands) you can get more info by selecting or highlighting the command, and pressing the 'F1' key on your keyboard. This will open up a detailed description in the command helper on the right side of your screen. It will display more information about the parameters required, and a more detailed description of what the command is and how it works.

[gasser_y]

@Jörg Vogel Thank you for your valuable tips.

[gasser_y]

Great. I will give it a try, hope to reach out for something.

Thank you so much.

[philboboADSK]

"And, what can be the alternative for 'largest extreme value' and 'smallest extreme value' distributions ??"

That depends on what those terms mean. Where did you get the information that your workstation follows such a distribution?

The same statistical probability distribution may have many different names depending on the context.

Those terms aren't specific to a specific statistical probability distribution without looking at their definition in the context of where you got that information.

For example, what MiniTab calls the "smallest extreme value" and "largest extreme value" distributions are described by Stat::Fit as Extreme Value Type 1A and Extreme Value Type 1B (see Stat::Fit manual, pages 126-129).

In FlexSim, you can use the extremevalue1a() and extremevalue1b() distribution functions to sample from those distributions as defined by Stat::Fit, which appear to be the same as the distributions defined by MiniTab as "smallest extreme value" (extremevalue1a) and "largest extreme value" (extremevalue1b).

If you didn't get those terms from MiniTab, then you need to look at the documentation of wherever you got those terms from to determine what actual distributions those refer to. They may actually be Gumbel (aka Generalized Extreme Value distribution Type-I), Fréchet (aka inverse weibull), or Weibull distributions.

[lou_keller]

Hi Gasser, Please purchase and add to your library, two very important books. First, "Statistical Distributions, 4th Edition," by Evans, Hastings and Peacock, published by Wiley Press. They explain every aspect of every distribution you've mentioned and a whole lot more. And second, "100 Statistical Tests," by G. Kanji, published by Sage. In the interim, If you have any specific questions regarding any of the distributions, let me know and I'll fill you in! I used to teach under-graduate and graduate statistics so the discussion should be fun! My best, Lou

Oh, here are the URLs you need:

http://www.wiley.com/WileyCDA/WileyTitle/productCd-1118097823.html

https://books.google.com/books/about/100_Statistical_Tests.html?id=M-_WDH-06voC

[mischa_spelt]

"Unfortunately I'm not a statistician"

Said no one ever 🙂