This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 1 Stat 430/Math468 – Notes #8 Chapter 4 The Multivariate Normal Distribution (Continued) Result: (Conditional distribution) Suppose ~ ( , ) p N X . Make the following partitioning 1 2 = X X X , 1 2 = , and 11 12 21 22 = . Assume that 22   . Then the conditional distribution of 1 X given 2 2 = X x is still normal with Mean = 1 1 12 22 2 2 ( ) + x and VarianceCovariance Matrix= 1 11 12 22 21 . Notation: Suppose the dimension of 1 X is 1 q × . Then the dimension of 2 X is p q . The result above says that 1 2 2 1 2 11 2  ( ) ~ ( , ) q N = X X x i i , where 1 1 2 1 12 22 2 2 ( ) = + x i and 1 11 2 11 12 22 21 = i . Example: (Bivariate Normal) Suppose 2 ~ ( , ) N X . Then 1 2 2 1 2 11 2  ( ) ~ ( , ) X X x N μ σ = i i , where 12 1 2 1 2 2 22 ( ) x σ μ μ μ σ = + i and 2 12 11 2 11 22 σ σ σ σ = i . Result: Suppose ~ ( , ) p N X . Then 1. 1 2 ( )' ( ) ~ p χ X  X  2. 1 2 {( ) ' ( ) ( )} 1 p P χ α α ≤ =  X  X  , where 2 ( ) p χ α is the upper (100 ) α th percentile of the 2 p χ distribution. Proof: is positive definite, then 1 is also positive definite and 1 1 2 2 1 = ....
View
Full
Document
This note was uploaded on 03/07/2010 for the course STAT 430 taught by Professor Hua during the Spring '10 term at University of Illinois at Urbana–Champaign.
 Spring '10
 Hua
 Statistics, Normal Distribution

Click to edit the document details