Dr. Shalabh. Indian Institute of Technology Kanpur

Similar documents
Special Instructions / Useful Data

Introduction to Matrices and Matrix Approach to Simple Linear Regression

X X X E[ ] E X E X. is the ()m n where the ( i,)th. j element is the mean of the ( i,)th., then

( ) = ( ) ( ) Chapter 13 Asymptotic Theory and Stochastic Regressors. Stochastic regressors model

Multiple Linear Regression Analysis

Qualifying Exam Statistical Theory Problem Solutions August 2005

Chapter 3 Multiple Linear Regression Model

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

Chapter 4 Multiple Random Variables

Lecture 3 Probability review (cont d)

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

ρ < 1 be five real numbers. The

Continuous Distributions

Training Sample Model: Given n observations, [[( Yi, x i the sample model can be expressed as (1) where, zero and variance σ

Lecture Note to Rice Chapter 8

Overview. Basic concepts of Bayesian learning. Most probable model given data Coin tosses Linear regression Logistic regression

X ε ) = 0, or equivalently, lim

ENGI 4421 Joint Probability Distributions Page Joint Probability Distributions [Navidi sections 2.5 and 2.6; Devore sections

LINEAR REGRESSION ANALYSIS

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

Lecture Notes 2. The ability to manipulate matrices is critical in economics.

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model

Module 7. Lecture 7: Statistical parameter estimation

Chapter 14 Logistic Regression Models

STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. x, where. = y - ˆ " 1

Chapter 5 Properties of a Random Sample

Parameter, Statistic and Random Samples

CHAPTER VI Statistical Analysis of Experimental Data

1 Solution to Problem 6.40

UNIVERSITY OF OSLO DEPARTMENT OF ECONOMICS

Summary of the lecture in Biostatistics

TESTS BASED ON MAXIMUM LIKELIHOOD

Chapter 3 Experimental Design Models

ENGI 3423 Simple Linear Regression Page 12-01

Multivariate Transformation of Variables and Maximum Likelihood Estimation

Department of Mathematics UNIVERSITY OF OSLO. FORMULAS FOR STK4040 (version 1, September 12th, 2011) A - Vectors and matrices

Class 13,14 June 17, 19, 2015

Some Different Perspectives on Linear Least Squares

The number of observed cases The number of parameters. ith case of the dichotomous dependent variable. the ith case of the jth parameter

Section 2 Notes. Elizabeth Stone and Charles Wang. January 15, Expectation and Conditional Expectation of a Random Variable.

22 Nonparametric Methods.

THE ROYAL STATISTICAL SOCIETY 2016 EXAMINATIONS SOLUTIONS HIGHER CERTIFICATE MODULE 5

Law of Large Numbers

Simulation Output Analysis

Functions of Random Variables

IFYMB002 Mathematics Business Appendix C Formula Booklet

4 Inner Product Spaces

Lecture Notes Types of economic variables

Chapter 5 Properties of a Random Sample

Quiz 1- Linear Regression Analysis (Based on Lectures 1-14)

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity

Simple Linear Regression Analysis

Linear Regression with One Regressor

Assignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class)

The Mathematical Appendix

Econometric Methods. Review of Estimation

Wu-Hausman Test: But if X and ε are independent, βˆ. ECON 324 Page 1

Chapter 4 Multiple Random Variables

= 2. Statistic - function that doesn't depend on any of the known parameters; examples:

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions.

Chapter 9 Jordan Block Matrices

Chapter 13, Part A Analysis of Variance and Experimental Design. Introduction to Analysis of Variance. Introduction to Analysis of Variance

Chapter 11 The Analysis of Variance

Recall MLR 5 Homskedasticity error u has the same variance given any values of the explanatory variables Var(u x1,...,xk) = 2 or E(UU ) = 2 I

Numerical Analysis Formulae Booklet

Econometrics. 3) Statistical properties of the OLS estimator

Maps on Triangular Matrix Algebras

The expected value of a sum of random variables,, is the sum of the expected values:

Mean is only appropriate for interval or ratio scales, not ordinal or nominal.

hp calculators HP 30S Statistics Averages and Standard Deviations Average and Standard Deviation Practice Finding Averages and Standard Deviations

Random Variables and Probability Distributions

Introduction to Probability

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions

STK3100 and STK4100 Autumn 2018

Application of Legendre Bernstein basis transformations to degree elevation and degree reduction

STATISTICAL INFERENCE

Probability and. Lecture 13: and Correlation

4. Standard Regression Model and Spatial Dependence Tests

Lecture 3. Sampling, sampling distributions, and parameter estimation

Simple Linear Regression and Correlation.

Midterm Exam 1, section 1 (Solution) Thursday, February hour, 15 minutes

STK3100 and STK4100 Autumn 2017

Detection and Estimation Theory

Line Fitting and Regression

ECON 5360 Class Notes GMM

Matrix Algebra Tutorial With Examples in Matlab

COV. Violation of constant variance of ε i s but they are still independent. The error term (ε) is said to be heteroscedastic.

MATH 247/Winter Notes on the adjoint and on normal operators.

TESTING FOR ORDER RESTRICTION ON MEAN VECTORS OF MULTIVARIATE NORMAL POPULATIONS

BAYESIAN INFERENCES FOR TWO PARAMETER WEIBULL DISTRIBUTION

ESS Line Fitting

DISTURBANCE TERMS. is a scalar and x i

Analysis of System Performance IN2072 Chapter 5 Analysis of Non Markov Systems

1. The weight of six Golden Retrievers is 66, 61, 70, 67, 92 and 66 pounds. The weight of six Labrador Retrievers is 54, 60, 72, 78, 84 and 67.

Complete Convergence and Some Maximal Inequalities for Weighted Sums of Random Variables

Discrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b

Point Estimation: definition of estimators

Maximum Likelihood Estimation

Arithmetic Mean Suppose there is only a finite number N of items in the system of interest. Then the population arithmetic mean is

Lecture 1 Review of Fundamental Statistical Concepts

Transcription:

Aalyss of Varace ad Desg of Expermets-I MODULE -I LECTURE - SOME RESULTS ON LINEAR ALGEBRA, MATRIX THEORY AND DISTRIBUTIONS Dr. Shalabh Departmet t of Mathematcs t ad Statstcs t t Ida Isttute of Techology Kapur

Quadratc forms If A s a gve matrx of order m ad X ad Y are two gve vectors of order m ad respectvely, the the quadratc form s gve by X ' AY m = = j= a x y j j where a ' s are the ostochastc elemets of A. j If A s square matrx of order m ad X = Y, the X ' AX = a x +... + a x + ( a + a ) x x +... + ( a + a ) x x. mm m m, m m, m m m If A s symmetrc also, the X ' AX = a x +... + a x + a x x +... + a x x mm m m, m m m m = axx j j = j= s called a quadratc form m varables x, x,, x m or a quadratc form X. To every quadratc form correspods a symmetrc matrx ad vce versa. The matrx A s called the matrx of quadratc form. The quadratc form X ' AX ad the matrx A of the form s called Postve defte f X ' AX > 0 for all x 0. Postve sem defte f X ' AX 0 for all x 0. Negatve defte f X ' AX < 0 for all x 0. Negatve sem defte f X ' AX 0 for all x 0.

3 If A s postve sem defte matrx the a 0 ad f a = 0 the a = 0 for all j, ad a j = 0 for all j. If P s ay osgular matrx ad A s ay postve defte matrx (or postve sem-defte matrx) the P' AP s also a postve defte matrx (or postve sem-defte matrx). A matrx A s postve defte f ad oly f there exsts a o-sgular matrx P such that A= P' P. A postve defte matrx s a osgular matrx. If A s m matrx ad ra ( A) = m < the AA' s postve defte ad A' A s postve semdefte. If A m matrx ad ra ( A) = < m <, the both AA ' ad AA' are postve semdefte. ( ) j

4 Smultaeous lear equatos The set of m lear equatos uows x, x,..., x ad scalars aj a x + a x +... + a x = b a x + a x +... + a x = b a x a x a x b m + m +... + m = m ad b, =,,..., m, j =,,..., of the form ca be formulated as AX = b where A s a real matrx of ow scalars of order vector of ow scalars gve by m called as coeffcet matrx, X s real vector ad b s real a a... a a a... a A=, s a m am am... am real matrx called as coeffcet matrx, x b x b =, s a vector of varables ad = s a real vector. x b m X b m

5 If A s osgular matrx, the AX = b has a uque soluto. Let B = [A, b] s a augmeted matrx. A soluto to AX = b exst tff ad oly f ra(a) = ra(b). If A s a m matrx of ra m, the AX = b has a soluto. Lear homogeeous system AX = 0 has a soluto other tha X = 0 f ad oly f ra (A) <. If AX = b s cosstet the AX = b has a uque soluto f ad oly f ra (A) = If a s the th dagoal elemet of a orthogoal matrx, the. Let the matrx be parttoed as A= [ a where s a vector of the elemets of th, a,..., a ] a colum of A. A ecessary ad suffcet codto that A s a orthogoal matrx s gve by the followg: ' ( ) = =,,..., aa for ' ( ) aa j = 0 for j =,,...,. a Orthogoal matrx A square matrx A s called a orthogoal matrx f A orthogoal matrx s o-sgular. If A s orthogoal, the AA' s also orthogoal. A ' A = AA' = I or equvaletly f A = A'. If A s a matrx ad let P s a orthogoal matrx, the the determats of A ad PAP ' are the same.

6 Radom vectors Let Y be radom varables the s called a radom vector., Y,..., Y Y = ( Y, Y,..., Y )' The mea vector of Y s EY = EY EY EY ( ) (( ( ), ( ),..., ( )) '. The covarace matrx or dsperso matrx of Y s Var( Y ) Var( Y) Cov( Y, Y)... Cov( Y, Y ) Cov( Y, Y ) Var( Y )... Cov( Y, Y ) Cov( Y, Y) Cov( Y, Y)... Var( Y) = whch s a symmetrc matrx. If Y, Y,..., Y are par-wse ucorrelated, the the covarace matrx s a dagoal matrx. If Var( Y ) = σ for all =,,, the Var Y I ( ) = σ.

7 Lear fucto of radom varable If Y, Y,..., Y are radom varables ad,,.., are scalars, the Y s called a lear fucto of radom varables Y, Y,..., Y. = If Y = ( Y, Y,..., Y)', K = (,,..., )' the K ' Y = Y, = the mea K ' Y s E( KY ' ) KEY ' ( ) EY ( ) ad the varace of s ' = = = K Y Var ( K Y ) = ( ) ' K ' Var Y K. Multvarate t ormal ldstrbuto tb t A radom vector Y = ( Y, Y,..., Y )' has a multvarate ormal dstrbuto wth mea vector μ = ( μ, μ,..., μ ) ad dsperso matrx Σ f ts probablty desty fucto s f Y Y Y ( π ) Σ ( μ, Σ ) = exp ( μ)' Σ ( μ) / / assumg Σ s a osgular matrx.

Ch-square dstrbuto 8 If Y, Y,...,, Y are detcally ad depedetly dstrbuted radom varables followg the ormal dstrbuto wth commo mea 0 ad commo varace, the the dstrbuto of Y s called the χ - dstrbuto wth degrees of freedom. = The probablty desty fucto of χ - dstrbuto wth degrees of freedom s gve as x f ( x ) = x exp ; 0 x. χ / < < Γ( /) If Y, Y,..., Y are depedetly dstrbuted followg the ormal dstrbuto wth commo meas 0 ad commo = varace σ, the has - dstrbuto wth degrees of freedom. Y χ σ If the radom varables Y, Y,..., Y are ormally dstrbuted wth o-ull meas μ, μ,..., μ but commo varace YY, the the dstrbuto b t of has o-cetral χ - dstrbuto b t wth degrees of freedom ad o-cetralty parameter λ = μ = = If Y, Y,..., Y are depedetly dstrbuted followg the ormal dstrbuto wth meas μ, μ,..., μ but commo varace σ the Y has o-cetral χ -dstrbuto wth degrees of freedom ad ocetralty parameter λ = μ. σ = σ =

9 If U has a Ch-square dstrbuto wth degrees of freedom the EU ( ) = ad Var( U ) =. If U has a ocetral Ch-square dstrbuto wth degrees of freedom ad ocetralty parameter λ the ( ) EU = + λ ad Var( U ) = + 4 λ. If U, U,..., U are depedetly dstrbuted radom varables wth each U havg a ocetral Ch-square dstrbuto wth degrees of freedom ad o cetralty parameter λ, =,,..., the U has ocetral Ch-square dstrbuto b t wth degrees of freedom ad ocetralty parameter = λ. = = = μ Σ. Let X ( X, X,..., X )' has a multvarate dstrbuto wth mea vector ad postve defte covarace matrx The X ' AX s dstrbuted as ocetral wth degrees of freedom f ad oly f s a dempotet matrx of ra. χ ΣA Let X = ( X, X,..., X ) has a multvarate ormal dstrbuto wth mea vector μ ad postve defte covarace matrx Σ. Let the two quadratc forms- χ X ' AX s dstrbuted b t d as wth degrees of ffreedom ad ocetralty parameter μ ' A μ ad X ' AX s dstrbuted as wth degrees of freedom ad ocetralty parameter χ μ A The X ' AX ad X' AX are depedetly dstrbuted f AΣ A = 0. ' μ.

0 t- dstrbuto If X has a ormal dstrbuto wth mea 0 ad varace, Y has a χ dstrbuto wth degrees of freedom, ad X ad Y are depedet radom varables, the the dstrbuto of the statstc T = The probablty desty fucto of T s X Y / s called the t-dstrbuto wth degrees of freedom. + + Γ t ft () t = ; - t. + < < Γ π X If the mea of X s o zero the the dstrbuto of Y / s called the ocetral t - dstrbuto wth degrees of freedom ad ocetralty parameter μ.

F- dstrbuto If X ad Y are depedet radom varables wth χ - dstrbuto wth m ad degrees of freedom respectvely, the the dstrbuto of the statstc probablty desty fucto of F s F = X / m Y / s called the F-dstrbuto wth m ad degrees of freedom. The m/ m+ m m+ Γ m m ff ( f) f = f ; 0 f. m + < < Γ Γ If X has a ocetral Ch-square dstrbuto wth m degrees of freedom ad ocetralty parameter λ; Y has a dstrbuto wth degrees of freedom, ad X ad Y are depedet radom varables, the the dstrbuto of X / m F = Y / s the ocetral F dstrbuto wth m ad degrees of freedom ad ocetralty parameter λ. χ