|
|
| Author |
Message |
Polvere@Lab science forum beginner
Joined: 27 Sep 2005
Posts: 1
|
Posted: Tue Sep 27, 2005 1:37 pm Post subject:
Which single descriptor for a probability distribution?
|
|
|
Hi everybody,
I hope this is the correct NG to ask to (and sorry for my english :)
I'm doing my thesis and I'd need a single descriptor that can represent in
the best way a probability distribution, something like a multimodal
gaussian one (actually, it would be an histogram, but you can image it like
that). I could use median, mode and variance, but I know that they are not
so representative, and, though, as I said, I need a single descriptor.
I was thining to consider that distribution as a figure, and to calculate
its mass center (which probably it not so good when you have a multimodal
distribution), or, better, some kinf of moment.
What do you think about it?
If you are aware of something better, I'm glad to learn from you
Thank you very much!
Ciao
Davide |
|
| Back to top |
|
 |
Rusty science forum beginner
Joined: 07 Sep 2005
Posts: 46
|
Posted: Wed Sep 28, 2005 7:44 pm Post subject:
Re: Which single descriptor for a probability distribution?
|
|
|
"Polvere@Lab" <dmonnetREMOVE@firenze.net> wrote in message
news:dhbp1u$peq$1@area.cu.mi.it...
| Quote: | Hi everybody,
I hope this is the correct NG to ask to (and sorry for my english :)
I'm doing my thesis and I'd need a single descriptor that can represent in
the best way a probability distribution, something like a multimodal
gaussian one (actually, it would be an histogram, but you can image it
like
that). I could use median, mode and variance, but I know that they are not
so representative, and, though, as I said, I need a single descriptor.
I was thining to consider that distribution as a figure, and to calculate
its mass center (which probably it not so good when you have a multimodal
distribution), or, better, some kinf of moment.
What do you think about it?
If you are aware of something better, I'm glad to learn from you
|
Distributions defined by a single parameter would be +v for a start and have
some simple shape. I would suggest a rectangular distribution on the
interval [0,A] or an exponential: p(x) = A exp(-Ax). Anything that is
symmetric requires both a mean and a variance at minimum which is two
parameters.
rusty |
|
| Back to top |
|
 |
Google
|
|
| Back to top |
|
 |
|
|
The time now is Fri Jan 09, 2009 9:13 pm | All times are GMT
|
|
Debt Consolidation | Debt Consolidation | MPAA | Bad Credit Mortgages | Mortgage Calculator
|
|
Copyright © 2004-2005 DeniX Solutions SRL
|
|
Other DeniX Solutions sites:
Electronics forum |
Medicine forum |
Unix/Linux blog |
Unix/Linux documentation |
Unix/Linux forums
|
Powered by phpBB © 2001, 2005 phpBB Group
|
|