s2 OCR how did u find it??

3 pi is exactly

I Used the binomial that way too, cant remember the answer though, was it something like 0.2 or 0.3?

x=2 was 0.195, x=1 was 0.371, and i think x=0 was 0.34~, so I got about...

0.094. About. And yes, I did just use Windows calculator :P

Reply 62

Urdegish

I personally approximated them to a binomial distribution, with N 50 and P 0.021. Then did 1 - p(x = 1,2,3).

For a binomial np needs to be bigger than 5, which it wasn't, it was 1.05.

So i think the only approximation is Po-(1.05)

Reply 63

I didnt think to try Poisson, on the strict rule of fact that the question stated it was a poisson distributed and part B) specified you had to use an approximation. And since I used Normal in part A), the rest followed... :smile:

Reply 64

pi is exactly 3

abc123

For a binomial np needs to be bigger than 5, which it wasn't, it was 1.05.

So i think the only approximation is Po-(1.05)

I thought that np had to be greater than 5 if you were using a normal approx to the binomial, not the other way round? Or am I completely wrong?

Reply 65

Urdegish

I personally approximated them to a binomial distribution, with N 50 and P 0.021. Then did 1 - p(x = 1,2,3).

Then model by a Poisson.

Reply 66

Urdegish

lol i'm confused now. In part a) we were given a poisson distribution, with a large lambda, so we could approximate using normal.

then in part B) we were given a binomial distributon (with a (p) from part (a) ), and as N was large, and P small, we could approximate using poisson...

I think thats how it was done

Reply 67

3 pi is exactly

I thought that np had to be greater than 5 if you were using a normal approx to the binomial, not the other way round? Or am I completely wrong?

Correct. That also. And np has to be less than 5 for Poisson to Normal. Looked at a question containing that before I entered exam.

Reply 68

abc123

I wish somebody would scan the paper, so we can remeber.

Reply 69

I think that part B was still in a Poisson distribution context. with lambda = 38x50 = 1900. Too big for Normal, and too big for Poisson, therefore have to use Binomial.

38x50 comes from 38 uniform events a week, and since there are 50 weeks then lambda increases uniformly

Reply 70

I also apologise if I appear to be throwing my proverbial weight around here, being a new member n' all that. :redface:

Reply 71

pi is exactly 3

But why approximate using a poisson when the original is a poisson? And do the poisson tables give enough accuracy and go high enough for when n is 50 and lamda is 1.05?

Reply 72

Ahh yes!! I remember now! Max n value was 30 in the formula book :smile:

Reply 73

yan2004

u had a binomial with X~B(50,0.021) which can be approximated to X~Po(1.05)

then the question asked for P(X>2) which is 1-P(X<=2) and u have to use the poisson formula. 1-(P(X=0)+P(X=1)+P(X=2))

Reply 74

Urdegish

Although that makes sense, and i had that thought in the exam, the reason why i doubt it is because there was a previous question on the paper (ques 5 i think) were we had to use a muiltiple of a Lambda 2, i don't think they would have ask us to use the same theory twice...

Reply 75

pi is exactly 3

yan2004

u had a binomial with X~B(50,0.021) which can be approximated to X~Po(1.05)

then the question asked for P(X>2) which is 1-P(X<=2) and u have to use the poisson formula. 1-(P(X=0)+P(X=1)+P(X=0))

Yeah i see where you're coming from but i still dont get why theyd want u 2 approximate using poisson when the thing was a poisson to begin with. The binomial works fine the same way and gets rid of any extra weirdness from another approx

Reply 76

Unfortunately, its a maths paper, and they are out to get us and make us trip up! :tongue:

Reply 77

yan2004

u had a binomial with X~B(50,0.021) which can be approximated to X~Po(1.05)

then the question asked for P(X>2) which is 1-P(X<=2) and u have to use the poisson formula. 1-(P(X=0)+P(X=1)+P(X=2))

Agreed.

Reply 78

3 pi is exactly

well part a) was poisson in context, but part b) was binomial in context....so if you look at it that way, it may not seem so wierd

Reply 79