S2 January 2015 IAL Q1) and Q7)

Watch this thread

Announcements

No university offers? Chat with other students about Ucas extra here >>

start new discussion reply

Page 1 of 1

Go to first unread

Skip to page:

φM23

Badges: 7

Rep:

Report Thread starter 6 years ago

1) For 1(b), Why isn't it P(X=1)? Also, I understand it is 2 days to the mean is doubled, but couldn't using X~Po(1.6) also include exactly 2 days? So why is this correct for less than 2 days?

2) For 1(c), please could someone explain this step by step?

3) For 7), where does Var(Y)=(4n/25) come from?

Probability is my worst topic and this is really bugging me.

Thanks for any help.

Attached files

4343.PNG (81.0 KB)
444.PNG (21.5 KB)
111111.PNG (53.4 KB)
87777.PNG (21.2 KB)

ghostwalker

Badges: 17

Report 6 years ago

(Original post by PhyM23)
1) For 1(b), Why isn't it P(X=1)?

X=1 is the probability that only one car was caught in the first 48 hours. There could be any number >= 1, and it still satisfies the criterion of catching a car within 48 hours.

Also, I understand it is 2 days to the mean is doubled, but couldn't using X~Po(1.6) also include exactly 2 days? So why is this correct for less than 2 days?

The parameter lambda is a continuous function (multiple) of the number of days.
What would lambda be if the distribution was for 2 days minus 1 hour?
Or two days minus 1 minute? Or minus 1 second?, Etc.

Also, although the Poisson is a discrete distribution, it's modelling a continous time period. What's the probability of a car arriving at exactly 2 days? Zero!

Don't feel I've really dealt with that bit properly, but hope it helps.

2) For 1(c), please could someone explain this step by step?

Conditional probabilty.

P(1 caught one day 1 AND 3 caught on day 2 | 4 caught over 2 days.)

The first two parts (red) refer to a single day period - hence lambda = 0.8.
Last past (green) refers to a 2 day period - hence lambda = 1.6.

Then standard conditional probability.

= P(1 caught one day 1 AND 3 caught on day 2 ) / P( 4 caught over 2 days.)

The two days are independent, so

= P(1 caught one day 1) P( 3 caught on day 2 ) / P( 4 caught over 2 days.)

3) For 7), where does Var(Y)=(4n/25) come from?

Probability is my worst topic and this is really bugging me.

Thanks for any help.

The binomial, which you are approximating, has a variance of npq, or np(1-p)
Hence n x (1/5) x (4/5) = 4n/25.

φM23

Badges: 7

Rep:

Report Thread starter 6 years ago

(Original post by ghostwalker)
X=1 is the probability that only one car was caught in the first 48 hours. There could be any number >= 1, and it still satisfies the criterion of catching a car within 48 hours.

The parameter lambda is a continuous function (multiple) of the number of days.
What would lambda be if the distribution was for 2 days minus 1 hour?
Or two days minus 1 minute? Or minus 1 second?, Etc.

Also, although the Poisson is a discrete distribution, it's modelling a continous time period. What's the probability of a car arriving at exactly 2 days? Zero!

Don't feel I've really dealt with that bit properly, but hope it helps.

Conditional probabilty.

P(1 caught one day 1 AND 3 caught on day 2 | 4 caught over 2 days.)

The first two parts (red) refer to a single day period - hence lambda = 0.8.
Last past (green) refers to a 2 day period - hence lambda = 1.6.

Then standard conditional probability.

= P(1 caught one day 1 AND 3 caught on day 2 ) / P( 4 caught over 2 days.)

The two days are independent, so

= P(1 caught one day 1) P( 3 caught on day 2 ) / P( 4 caught over 2 days.)

The binomial, which you are approximating, has a variance of npq, or np(1-p)
Hence n x (1/5) x (4/5) = 4n/25.

Thank you very much for such a detailed reply.

1) This makes sense and your explanation of the second part is exactly what I was looking for.

2) This makes more sense, but why isn't P(4 caught over 2 days) in the numerator too?

3) Ah yes I should have spotted that :facepalm:

Zacken

Badges: 22

Rep:

Report 6 years ago

(Original post by PhyM23)
2) This makes more sense, but why isn't P(4 caught over 2 days) in the numerator too?

Look up conditional probability in the S1 section of your formula booklet:

$P(A | B) = \frac{P(A \cap B)}{P(B)}$

φM23

Badges: 7

Rep:

Report Thread starter 6 years ago

(Original post by Zacken)
Look up conditional probability in the S1 section of your formula booklet:

$P(A | B) = \frac{P(A \cap B)}{P(B)}$

That's what's confusing me.

Let P(1 caught on day 1) = P(A)
P(3 caught on day 2) = P(B)
P(4 caught over 2 days) = P(C)

P(A and B|C) = P(A and B and C) / P(C)

Why isn't it this?

Zacken

Badges: 22

Rep:

Report 6 years ago

(Original post by PhyM23)
That's what's confusing me.

Let P(1 caught on day 1) = P(A)
P(3 caught on day 2) = P(B)
P(4 caught over 2 days) = P(C)

P(A and B|C) = P(A and B and C) / P(C)

Why isn't it this?

Ah, I see what you mean.

It's just that A and B and C is (1 caught on day 1 and 3 caught on day 2 and 4 caught over 2 days).

But (1 caught on day 1 and 3 caught on day 2) is already (4 caught over 2 days). So $A \cap B \cap C = A \cap B$ .

Kind of like $\{1, 2, 3\} \cap \{3, 4, 5\} \cap \{1, 2, 3, 4, 5, 6, 7,...\} = \{1, 2, 3\} \cap \{3, 4, 5\}$

Edit: see below, ghostwalker explained it better.

ghostwalker

Badges: 17

Report 6 years ago

You're quite right. I missed a step out.

P(A and B|C) = P(A and B and C) / P(C)

The event "A and B" (i.e. 1 on day 1, and 3 on day 2), implies 4 over 2 days, i.e. C.

So, the event "A and B" is a subset of event "C".

$A\cap B \subseteq C$

Hence P(A and B and C) = P(A and B).

Edit: ninja'd

φM23

Badges: 7

Rep:

Report Thread starter 6 years ago

(Original post by Zacken)
Ah, I see what you mean.

It's just that A and B and C is (1 caught on day 1 and 3 caught on day 2 and 4 caught over 2 days).

But (1 caught on day 1 and 3 caught on day 2) is already (4 caught over 2 days). So $A \cap B \cap C = A \cap B$ .

Kind of like $\{1, 2, 3\} \cap \{3, 4, 5\} \cap \{1, 2, 3, 4, 5, 6, 7,...\} = \{1, 2, 3\} \cap \{3, 4, 5\}$

Edit: see below, ghostwalker explained it better.

(Original post by ghostwalker)
You're quite right. I missed a step out.

P(A and B|C) = P(A and B and C) / P(C)

The event "A and B" (i.e. 1 on day 1, and 3 on day 2), implies 4 over 2 days, i.e. C.

So, the event "A and B" is a subset of event "C".

$A\cap B \subseteq C$

Hence P(A and B and C) = P(A and B).

Edit: ninja'd

Haha