Re: Probability of an "Identical And Wrong" result
- From: "D. C." <enharmonix@xxxxxxxxx>
- Date: 10 Sep 2006 11:23:40 -0700
Mike wrote:
[snip]
If the probability of an incorrect answer is .01 from both algorithms
for a given datum the joint probability that both will generate an
error is the product; .01 * .01 = .0001. But this includes the case
where the answers differ. I need to understand the probability that
the answers are identical and wrong.
Any help or pointers to discussions of this problem would be most
appreciated.
--Thank you,
--Mike Jr
Step 1: http://mathworld.wolfram.com/ProbabilityAxioms.html and
http://mathworld.wolfram.com/Probability.html (note Conditional Events,
may make a difference how you represent your probabilities)
Step 2: This is a probability qusetion so cross post to sci.math and
alt.sci.math.probability if you haven't already.
Good news: easy to solve once defined. Bad news: hard to define. You
can't guess at how likely both are going to fail with identical results
unless you know how likely they are to _have_ identical results in the
first place (right or wrong). Once you know that, it is easy to solve.
Best case is simple algebra, worst case is a sigma function.
Like I said though, it's probability so check sci.math and
alt.sci.math.probability because they proabbly have some more tricks up
their sleeves.
Cheers and good luck,
D. C.
.
- References:
- Prev by Date: Re: help: a O(n) algorithm to find the longest path in a tree
- Next by Date: Re: Probability of an "Identical And Wrong" result
- Previous by thread: Re: Probability of an "Identical And Wrong" result
- Next by thread: Re: Probability of an "Identical And Wrong" result
- Index(es):
Relevant Pages
|
