The Student Room Group

S2 ocr mei 2016

Scroll to see replies

For 4b)
What was the mean, sum of x and the sum of x^2???

Posted from TSR Mobile
Original post by ROSS1998JONES2
What use 2 approximations, from Binomial to Poission and then Poission to Normal? Seems unnecessary for me and I'm not pretty sure they wont credit that on the mark schemes, never seen them credit that before. Unless you can give me an example otherwise?


They already made you use poisson distribution for when there was 500 of that DNA thing, so you had a mean of 6, all you had to do to approximate it to normal was to multiply 6 by 100.
Original post by Studious_Student
They already made you use poisson distribution for when there was 500 of that DNA thing, so you had a mean of 6, all you had to do to approximate it to normal was to multiply 6 by 100.


Tbf, there are two ways you can approximate this solely because of this reason.

Besides, there was a similar case for the 2015 paper where X~N(lambda, lambda) was allowed so it'll probably be allowed for this one as well

I did Bin ~> Norm but Poisson ~> Norm should be okay
Posted from TSR Mobile
Original post by Studious_Student
They already made you use poisson distribution for when there was 500 of that DNA thing, so you had a mean of 6, all you had to do to approximate it to normal was to multiply 6 by 100.


The actual distribution there though is binomial: X~B(50000,0.012), and hence by approximating X~N(600,600) you have approximated twice, first to poisson and the to Normal, and hence pretty sure you wont be credited I'm afraid
Original post by Leechayy
Tbf, there are two ways you can approximate this solely because of this reason.

Besides, there was a similar case for the 2015 paper where X~N(lambda, lambda) was allowed so it'll probably be allowed for this one as well

I did Bin ~> Norm but Poisson ~> Norm should be okay
Posted from TSR Mobile



Yes there are 2 ways, but one requires 2 approximations: the point of an approximation is that is different to the actual distribution. Hence you cannot approximate from the Poisson only because that is in itself an approximation
Screenshot_20160615-213347.jpg
Original post by ROSS1998JONES2
Yes there are 2 ways, but one requires 2 approximations: the point of an approximation is that is different to the actual distribution. Hence you cannot approximate from the Poisson only because that is in itself an approximation

This was from the 2015 paper where they asked you to find the Poisson distribution, then asked you to approximate to Normal just like with this paper. They accepted and allowed full credit here.
Original post by Studious_Student
Screenshot_20160615-213347.jpg
This was from the 2015 paper where they asked you to find the Poisson distribution, then asked you to approximate to Normal just like with this paper. They accepted and allowed full credit here.


Fair enough, I dont understand what it means by Normal approximation to Poisson, surely it should be normal approximation from poisson, but that does indeed give you full marks. But that is assuming 2 approximations. Cant guarantee the same applies this year however.
Original post by ROSS1998JONES2
Fair enough, I dont understand what it means by Normal approximation to Poisson, surely it should be normal approximation from poisson, but that does indeed give you full marks. But that is assuming 2 approximations. Cant guarantee the same applies this year however.


I hope it still does, I've already lost enough marks as it is. :frown:
For the last question (hypothesis testing for mean) i done everything right, but i completely ignored the variance/n in my normal distribution (just used the calculated variance) and also didn't work out the z value for the mean..........how many marks d'you think i can get? Right after the exam i realized what i forgot- i was rushed for time. It was an easy paper, so frustrated i messed up on that question.
What do you guys think the grade boundaries are going to be like?
Original post by Jasmine149
What do you guys think the grade boundaries are going to be like?


67 A*
63 A
59 B
54 C

Posted from TSR Mobile
Reply 31
Well
Reply 32
Original post by ROSS1998JONES2
Fair enough, I dont understand what it means by Normal approximation to Poisson, surely it should be normal approximation from poisson, but that does indeed give you full marks. But that is assuming 2 approximations. Cant guarantee the same applies this year however.

Well, you can't do that approximation from binomial to normal, because it only works when p is around 0.5 so that it's about symmetrical
Reply 33
Original post by Leechayy
Tbf, there are two ways you can approximate this solely because of this reason.

Besides, there was a similar case for the 2015 paper where X~N(lambda, lambda) was allowed so it'll probably be allowed for this one as well

I did Bin ~> Norm but Poisson ~> Norm should be okay
Posted from TSR Mobile

No, the only way is from poisson to normal, cuz normal can only be used for binomial when p is not near 0 or 1.
Reply 34
Original post by ROSS1998JONES2
This is what I can remember

Q1:s-smilie:catter diagram:
1 mark for labelled axes, 2 marks for correctly plotted points
Spearmans: -0.366
Spearmans hypothesis test: accept H0.
Pearsons not appropriate since not bivariate normal because not elliptical.
No assumptions required for Spearmans.
1% significance level is 1% probability of rejecting H0 when it is in fact true.

Q2:
Independent: mutations dont affect probability of another mutation, random:mutations are unpredictable
Binomial calculation: 20C1 x 0.012^1 x 0.988^19 or something like that
Approximation to posission: mean = 500x0.012, then use tables etc.
Approximation to normal, mean=50000x0.012, variance=50000x0.012x0.998 then standardise and use tables etc.

Q3:
Simple normal calculations
Manufacturer claims over 95% etc.: calculate value= 0.9502 so claim is true.
Find sigma and mu, sigma =6350 and mu = 60500 or 50850 or something like that?
Draw distributions: X has lower mean and variance so less spread but taller
Calculate value of h when 99.9%: I got 40500 or there abouts

Q4:
Calculate expected frequency of female and likes maths, show contribution=...
Chi test, add up contributions, accept H0.
Normal test, calculate mean and variance or standard distribution, calculate test statistic, reject H0.

If any one remembers anything else let me know.


For the question about assumptions for Spearmans rank. I wrote "have bivariate distribution"
You think Id get the mark?
Original post by JK11
For the question about assumptions for Spearmans rank. I wrote "have bivariate distribution"
You think Id get the mark?


No because bivariate distribution is not required for Spearman's rank.
I know this isn't the ocr mei c3 thread (I can't find it lol), but I got 15/18 in my coursework, is that a grade A or B?
Original post by VlAd x
No, the only way is from poisson to normal, cuz normal can only be used for binomial when p is not near 0 or 1.


Ahhh right. Thanks :redface:

Posted from TSR Mobile
Original post by Jasmine149
I know this isn't the ocr mei c3 thread (I can't find it lol), but I got 15/18 in my coursework, is that a grade A or B?


I think it's a very close A.
Original post by Leechayy
Ahhh right. Thanks :redface:

Posted from TSR Mobile


does this mean that youd use lambda for both the mean and the variance and then use the normal test? cos that what i did..?

Quick Reply

Latest