Probability - Conditional Expectation

  • #1
My professor explained this concept absolutely horribly and I have no idea how to do these problems.

Let A and B be independent Poisson random variables with parameters α and β, respectively. Find the conditional expectation of A given A + B = c.
(Hint: For discrete random variables, there is no conditional density. Use the definition of conditional probability.)

Attempt:
Starting with the definition, f(A | A + B = c) = [f(A, A+B=c)] / [f(A+B=c)]

Not sure how to proceed.
 

Answers and Replies

  • #2
22,129
3,298
We will have to work with A+B. Do you know the probability distribution of this??
 
  • #3
Ray Vickson
Science Advisor
Homework Helper
Dearly Missed
10,706
1,722
My professor explained this concept absolutely horribly and I have no idea how to do these problems.

Let A and B be independent Poisson random variables with parameters α and β, respectively. Find the conditional expectation of A given A + B = c.
(Hint: For discrete random variables, there is no conditional density. Use the definition of conditional probability.)

Attempt:
Starting with the definition, f(A | A + B = c) = [f(A, A+B=c)] / [f(A+B=c)]

Not sure how to proceed.

It is best to be clear and to use correct notation: you want P{A=k|A+B=c} for the possible values of k in {0,1,2,...}. So, you need to compute P{A=k & A+B=c} in the numerator (and, of course, you need P{A+B=c} in the denominator).

Do you know the distribution of A+B? It should be in your textbook or course notes; if not, look on-line, or work it out for yourself from first principles, using the distributions of A and B and the formula for the distribution of a sum of independent random variables (really: it is not that hard!).

RGV
 
Last edited:
  • #4
The distribution for a Poisson distribution is p(x) = [e^(-λ)*λ^x] / x!
 
  • #5
22,129
3,298
Yes, that is the distribution for A and B with [itex]\lambda=\alpha[/itex] and [itex]\lambda=\beta[/itex] respectively.

But we are asking about the distribution of A+B.
 
  • #6
A + B also has a Poisson distribution with parameters Poisson(A+B), as the sum of independent Poisson random variables has a Poisson distribution.
 
  • #7
22,129
3,298
OK, that's good. Now we want to figure out (for fixed c)

[tex]f(A=x~|~A+B=c)[/tex]

In order to find to, we want to find

[tex]f(A=x,~A+B=c)[/tex]

Of course, this is equal to

[tex]f(A=x,~B=c-x)[/tex]

Can you find this?? This is just a two-dimensional pmf. Remember that A and B are independent, so you can find it easily.

Then we also nee to find

[tex]f(A+B=c)[/tex]

This should be easy since you just figured out the distribution of A+B.
 
  • #8
Not sure how to go about finding f(A=x , B=c−x)
 
  • #9
Ray Vickson
Science Advisor
Homework Helper
Dearly Missed
10,706
1,722
Not sure how to go about finding f(A=x , B=c−x)

You told us that A and B are independent. What do you think that means?

RGV
 
  • #10
Can anyone show me this problem step by step? I'm not picking up on any of this question, which is why I posted this.
 
  • #11
Ray Vickson
Science Advisor
Homework Helper
Dearly Missed
10,706
1,722
Can anyone show me this problem step by step? I'm not picking up on any of this question, which is why I posted this.

No, we can't. That is not how this forum works. However, I will give you a hint: if your professor did not explain things to your satisfaction, and if, for some reason you do not have access to course notes or to a textbook, then *look online*. Google 'independent + probability' to turn up hundreds of articles at various levels of sophistication, from step-by-step explanations to abstract discussions.

RGV
 
Last edited:
  • #12
P(A|A+B=c)
= P(A|B=c-A)
= P(A and B=c-A) / P(B=c-A)

=P(α + β) / P(β) ?
 
  • #13
Ray Vickson
Science Advisor
Homework Helper
Dearly Missed
10,706
1,722
P(A|A+B=c)
= P(A|B=c-A)
= P(A and B=c-A) / P(B=c-A)

=P(α + β) / P(β) ?

I have no idea what you mean by P(α + β) or P(β). I know what α and β are, and I know what is meant by P(A=u) or P(B=v) and how to write them in terms of α, β, u and v, but I cannot figure out your P(α+β), etc. Anyway, I certainly would get something very different from what you wrote.

RGV
 

Related Threads on Probability - Conditional Expectation

  • Last Post
Replies
0
Views
909
  • Last Post
Replies
4
Views
908
Replies
1
Views
2K
Replies
5
Views
558
  • Last Post
Replies
6
Views
1K
  • Last Post
Replies
1
Views
2K
  • Last Post
Replies
2
Views
3K
  • Last Post
Replies
0
Views
1K
  • Last Post
Replies
0
Views
2K
  • Last Post
Replies
4
Views
2K
Top