View Full Version : Hypergeometric Distribution - Help needed to understand
amanbhai 23rd April 2005, 11:44 AM folks anyone know about statistical distribution that's called Hypergeometric distribution.
Anyone who who can help me understand its calculation, problem solving, etc.
well i found the theory & other stuff but i need someone's help in these things.
Govind 23rd April 2005, 11:57 AM Mr.Amanbhai,
Please post the threads in the right forum. This was posted in reading room. This is second time in a row you are posting in different forums. Posting in right forum help search easier and also gets the attention of subject matter experts who usually visit the forum.
Please review this electronic statistical handbook from NIST (http://www.itl.nist.gov/div898/handbook/index.htm).
Distributions like Normal, binomial, Poisson, Log normal, Exponential, Weibull are more frequently used in manufacturing and reliability.
Distributions like Gamma, Hypergeomeric are used in pure stats and Sampling research applications. Although these distributions may have wide variety of applications in manufacturing quality, I have not seen a wide usage so far.
Here is an excellent practical life example of hypergeometric distribution from the web:
Application of the Hypergeometric Distribution (http://www.amstat.org/sections/srms/Proceedings/y2001/Proceed/00037.pdf)
Regards,
Govind.
amanbhai 23rd April 2005, 01:06 PM I'm new here in elsmar, & the other thing is that I wanna learn more about elsmar & stuff. thankful to you if you give me some insight in that. :D i mean how to use forums, & use elsmar efficiently etc.etc
regards
Govind 23rd April 2005, 04:46 PM Mr.Amanbhai,
There is a topic exclusively available in Elsmar for new users.
http://elsmar.com/Forums/forumdisplay.php?f=86
Very informative. Please go through the threads to get an idea of the features available.
w.r.t your question,
Mr.Tim.F has provided a very detailed feedback for your same question in ASQ Discussion forum. Even Quality Council of Indiana primer has only a page and half and example (similar to the example provided Mr.Tim.F) in this topic.
From the exam point of view, it is merely formula substitution. However,It will be interesting to read the practical applications and understand so that you will find an opportunity to apply in real life situations.
Regards,
Govind.
Statistical Steven 23rd May 2005, 05:01 PM folks anyone know about statistical distribution that's called Hypergeometric distribution.
Anyone who who can help me understand its calculation, problem solving, etc.
well i found the theory & other stuff but i need someone's help in these things.
The hypergeometric is used to calculate the probability of an event for a sample taken from a finite population. That is the true rate for a given event is known in the population. If it is an infinite population or the true percent fo the event is not know, you use the binomial distribution for the sample. Hope this helps.
Tim Folkerts 23rd May 2005, 11:50 PM Steven,
Welcome to the Cove. I see you are jumping right in posting on topics about stats (which is where I usually post as well).
What a weird coincidence. One Steve takes a rare weekend off from posting (Steve Prevette - "a passionate statistician") and :magic: a new "Statistical Steven" pops up to fill in. We seem to have entered "The Twilight Cove" :lmao:
Tim F
Statistical Steven 24th May 2005, 10:39 AM Just trying to provide good statistical advice. Are there alot of statisticians here?
Jim Wynne 24th May 2005, 11:11 AM One Steve takes a rare weekend off from posting (Steve Prevette - "a passionate statistician") and http://elsmar.com/Forums/images/smilies/magic.gif a new "Statistical Steven" pops up to fill in. We seem to have entered "The Twilight Cove" http://elsmar.com/Forums/images/smilies/lmao.gif
Sounds like a call for Project Steve (http://www.ncseweb.org/article.asp?category=18)
Statistical Steven 24th May 2005, 11:30 AM Sounds like a call for Project Steve (http://www.ncseweb.org/article.asp?category=18)
Cute! :) :thanx: :lol: :lol: :lol:
Steve Prevette 24th May 2005, 05:34 PM I'm back . . . and I see we now have two statisticians named Steve.
Not quite as bad as a model railroad club I was a member of in Monterey CA. I believe I was "Steve 3" and even before I left the area there was a "Steve 4" and that in a group of only 20 people.
Let's see - now how would that work with a hypergeometric (so I can keep this on topic)? What is the probability of selecting two club members out of the 20 (of which 4 members are named Steve) and having exactly one of the two chosen members have the name "Steve"?
=HYPGEOMDIST(1,2,4,20)
=34%
Ron Rompen 25th May 2005, 06:28 PM Not being a statistician I won't even ATTEMPT to refute your answer, Steve 3 (if that is who you REALLY are).
However, your answer looks like an Excel function call, which doesn't match the information you gave.
2 picked, 1 chosen, 4 probables, 20 total
Would that not be written as
=HYPGEOMDIST (2,1,4,20) ?
(Haven't run this through Excel, so I don't know the answer...and I may very well be mistaken. As someone noted before, Hypo distributions are not widely used in industry (at least not to my knowledge))
Steve Prevette 25th May 2005, 06:33 PM However, your answer looks like an Excel function call, which doesn't match the information you gave.
2 picked, 1 chosen, 4 probables, 20 total
Would that not be written as
=HYPGEOMDIST (2,1,4,20) ?
The Hypergeometric values asked for in Excel are (using the function help for Hypgeomdist):
Sample_s - the number of successes in the sample (we are looking for there to be 1 Steve in the sample)
Number_sample - the size of the sample (2 members)
Population_s - the number of successes in the population (there are 4 Steve's in the club's population)
Number_pop - the size of the population (20 members)
So, the values in the parenthesis are (1,2,4,20)
Yes, Hypergeometric is rather specialized, but is useful when dealing with small populations (less than 100) and random sampling. Recently I had a real example where a sample was drawn, and they got what they thought was an unlikely result, so they wanted to backfit how likely it was to get that result.
|
|