From: Mike Harrison on
Hi, I've got a question about probability. I'm sure I managed to do
these at school, but that was many years ago!

Given an event with a 1 in 125 chance of occurring, what is the
probability of that event happening exactly 3 out of 8 times?

And also the probability of the event happening *at least* 3 out of 8
times?

If anyone could explain how this is calculated, that would be really
helpful. Many thanks!


From: Gerry Myerson on
In article
<3b00fdf9-4782-4e57-b35d-902d612be03a(a)e5g2000yqn.googlegroups.com>,
Mike Harrison <mjh626(a)yahoo.co.uk> wrote:

> Hi, I've got a question about probability. I'm sure I managed to do
> these at school, but that was many years ago!
>
> Given an event with a 1 in 125 chance of occurring, what is the
> probability of that event happening exactly 3 out of 8 times?

Assuming independence,
(8-choose-3) x (124-to-the-5th) / (125-to-the-8th).

8-choose-3 is (8 x 7 x 6) / (1 x 2 x 3) = 56.

> And also the probability of the event happening *at least* 3 out of 8
> times?

That's a little messier. It's
1 minus (probability of 0 plus probability of 1 plus probability of 2)
where those individual probabilities are calculated in a similar way
to the probability of 3 done earlier.

> If anyone could explain how this is calculated, that would be really
> helpful. Many thanks!

Pick up an intro probability textbook.

--
Gerry Myerson (gerry(a)maths.mq.edi.ai) (i -> u for email)
From: William Elliot on
On Thu, 15 Jul 2010, Mike Harrison wrote:

> Given an event with a 1 in 125 chance of occurring, what is the
> probability of that event happening exactly 3 out of 8 times?
>
> And also the probability of the event happening *at least* 3 out of 8
> times?
>
> If anyone could explain how this is calculated, that would be really
> helpful. Many thanks!

Let p be the probality of an event and q = 1 - p the probality of
the negation of the event. Then
1 = 1^n = (p + q)^n = sum(k=0,n) C(n,r) p^k q^(n-k)

The probablity of the event happening exactly k times in n tries
is p_k = C(n,r) p^k q^(n-k); the probablity of it happening
at least k times in n tries is sum(j=k,n) p_j; and the probablity
of it happening at most k times in n tries is sum(j=0,k) p_j.
From: porky_pig_jr on
On Jul 16, 1:42 am, Mike Harrison <mjh...(a)yahoo.co.uk> wrote:
> Hi, I've got a question about probability. I'm sure I managed to do
> these at school, but that was many years ago!
>
> Given an event with a 1 in 125 chance of occurring, what is the
> probability of that event happening exactly 3 out of 8 times?
>
> And also the probability of the event happening *at least* 3 out of 8
> times?
>
> If anyone could explain how this is calculated, that would be really
> helpful. Many thanks!

This is binomial distribution.

First, consider a single trial. Let p = probability of event occured
(won). p = 0.008.
q = 1 - p = probability of event did not occured (lost). q = 0.992. So
in a single trial we win
with probability p = 0.008, loose with probability q = 0.992.

Now. to win exactly 3 out of 8 times means 3 wins and 5 losses,
probability of that is
(p^3) (q^5) = (0.008^3)(0.992^5). This 3 out of 5 wins can happen in
8C3 ("8 choose 3") ways.
You can look up the formula. 8C3 = (8!) / (3! (8-3)!). So you have to
multiply by 8C3 and so the answer is
(8C3) (0.008^3) (0.992^5).

So, again, this is binomial distribution: winning exactly k out of n
trials, with p = probability of winning.

Now, to win at least 3 out of 8 means to win exactly 3 out of 8 OR
exactly 4 out of 8 blah blah blah or exactly 8 out of 8. All these
events are mutually disjoint. That is, you CANNOT win, say, exactly 4
*and* 7 out of 8 in the same trial. It's either 4 or 7. And in case of
a finite collection of mutually disjoint events, their total
probability is the sum of individual probabilities. That is,
P(3 out of 8 OR 4 out of 8 OR ... OR 8 out of 8) = P(3 out of 8) + P(4
out of 8) + ... + P(8 out of 8). Each is binomial distribution you
already know how to compute. Done.
From: Mike Harrison on
On Jul 16, 7:08 am, "porky_pig...(a)my-deja.com" <porky_pig...(a)my-
deja.com> wrote:

> Now. to win exactly 3 out of 8 times means 3 wins and 5 losses,
> probability of that is
> (p^3) (q^5) = (0.008^3)(0.992^5). This 3 out of 5 wins can happen in
> 8C3 ("8 choose 3") ways.
> You can look up the formula. 8C3 = (8!) / (3! (8-3)!). So you have to
> multiply by 8C3 and so the answer is
> (8C3) (0.008^3) (0.992^5).

Thanks. So I make that 2.754e-5 or 1-in-36306, is that right?

And the calculation for at least 3 out of 8 is:

(8C3) (0.008^3) (0.992^5) +
(8C4) (0.008^4) (0.992^4) +
(8C5) (0.008^5) (0.992^3) +
(8C6) (0.008^6) (0.992^2) +
(8C7) (0.008^7) (0.992^1) +
(8C8) (0.008^8) (0.992^0)

= 2.782e-5 or 1-in-35942.

Which seems hardly any more likely than the 3 out of 8 on its own. I
suppose that's because the chances of getting more than 3 quickly
become extremely small.

I hope I've calculated it right anyway.