Conditional Multinomial Problem

  • Context: Graduate 
  • Thread starter Thread starter FaradayLaws
  • Start date Start date
  • Tags Tags
    Conditional
Click For Summary
SUMMARY

The discussion centers on proving that the conditional distribution of Y1 given Y3=m, where Y1, Y2, Y3 follow a multinomial distribution with parameters (n, p1, p2, p3), is a binomial distribution with parameters (n-m, p1/(p1+p2)). The participants detail the steps involved in deriving this result, starting from the joint probability P(Y1=y1, Y3=m) and applying the multinomial distribution properties. The final simplification leads to the binomial form, confirming the conditional distribution as stated.

PREREQUISITES
  • Understanding of Multinomial Distribution
  • Knowledge of Conditional Probability
  • Familiarity with Binomial Distribution
  • Proficiency in Combinatorial Notation
NEXT STEPS
  • Study the properties of Multinomial Distribution in detail
  • Learn about Conditional Probability and its applications
  • Explore the derivation of Binomial Distribution from Multinomial Distribution
  • Practice combinatorial proofs and simplifications in probability theory
USEFUL FOR

Statisticians, data scientists, and mathematicians interested in advanced probability theory and its applications in statistical modeling.

FaradayLaws
Messages
8
Reaction score
0
If Y1,Y2,Y2 ~ Multinomial with parameter (n,p1,p2,p3)
Prove that the conditional distribution of Y1 given Y3=m (m<n)
is a binomial with ( (n-m), p1/(p1+p2) )

p1+p2+p3=1
y1+y2+y3=n
y3=m

My Attempt:
P( Y1=y1| y3=m) = P(Y1=y1, Y3=m)/ P(Y3=m)

( n choose m and y1 ) p1^y1*p3^m / ( n choose m) p3^m*(p1+p2)^n-m

leaving me with
(n-m)!/ y1! (p1/ p1+p2)^y1((p1+p2) ^y2))
...
can't seem to simplyfy this to become a binomial
honestly stuck here!

Thanks!
 
Physics news on Phys.org
(n choose m)(n-m choose y1) p1^y1*p2^(n-m-y1)*p3^m
divided by
(n choose m) p3^m*(p1+p2)^(n-m)

simplifies to

(n-m choose y1)p1^y1 p2^(n-m-y1) (p1 + p2)^(m - n). Rearrange to obtain the result.
 
Last edited:

Similar threads

  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 29 ·
Replies
29
Views
6K
Replies
6
Views
2K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 9 ·
Replies
9
Views
2K
Replies
11
Views
10K
  • · Replies 3 ·
Replies
3
Views
8K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K