Re: Normal distribution with non-negative distribution
- From: Richard Ulrich <Rich.Ulrich@xxxxxxxxxxx>
- Date: Sat, 02 Jun 2007 21:59:08 -0400
On Sat, 02 Jun 2007 14:18:40 -0700, royend@xxxxxxxxx wrote:
A normal distrubution where the mean is x and its standard deviation
is bigger than this (x + y), result in a normal distribution where a
percentage is below 0. In my case this is impossible, as the x-axis
corresponds to time and how long it takes to find a solution. And,
this can never be done before the problem is presented (at 0 seconds).
How may I present a correct distribution of the possibility to solve a
problem?
Example:
The following numbers correspond to a test to find out how long a
student needs to answer a particular question:
1, 15, 2, 2, 1
The equals a mean of 2.1 and a standard deviation at 6.06.
Building a normal distribution with this data gives a graph that
predicts that 25% of the students will answer this question before it
is asked (quite fantastic?).
Is there any other distribution model, or any way to use a normal
distribution in my case?
When you start with times that are seconds-to-Xs, you often
can get a better model when you express the numbers as Rates,
for example, "X-per-minute".
This improves the model for the above, and puts the
necessary truncation at the other end (which seems
easier to justify, here).
--
Rich Ulrich, wpilib@xxxxxxxx
http://www.pitt.edu/~wpilib/index.html
.
- References:
- Normal distribution with non-negative distribution
- From: royend
- Normal distribution with non-negative distribution
- Prev by Date: Normal distribution with non-negative distribution
- Next by Date: Re: Normal distribution with non-negative distribution
- Previous by thread: Normal distribution with non-negative distribution
- Next by thread: Re: Normal distribution with non-negative distribution
- Index(es):
Relevant Pages
|
|
