13 - People.stat.sfu.ca - Simon Fraser University
13 - People.stat.sfu.ca - Simon Fraser University
13 - People.stat.sfu.ca - Simon Fraser University
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Lecture <strong>13</strong>: Method of Moments<br />
To estimate p parameters solve<br />
¯X = E θ (X)<br />
X¯<br />
2 = E θ (X 2 )<br />
. .<br />
X¯<br />
p = E θ (X p )<br />
Essentially always consistent.<br />
Did Gamma(α,β) example.<br />
Used as starting points for Newton-Raphson.<br />
To solve g(θ) = 0 begin with initial value θ 0 and iteratively define<br />
( −1g(θ<br />
θ (k+1) = θ (k) − Dg(θ )) (k) (k) ).<br />
Here Dg is the p ×p matrix with i,jth<br />
∂g i (θ)<br />
∂θ j<br />
.<br />
Richard Lockhart (<strong>Simon</strong> <strong>Fraser</strong> <strong>University</strong>) STAT 801=830 Statisti<strong>ca</strong>l Inference Fall 2012 1 / 3
Lecture <strong>13</strong>: quadratic approximation to l<br />
Two term Taylor expansion of l about ˆθ:<br />
[ ]<br />
2 l(ˆθ)−l(θ) ≈ (θ− ˆθ) T I(θ 0 )(θ − ˆθ)<br />
where minus the second derivative matrix has been replaced by the<br />
Fisher information matrix.<br />
Worked on large sample theory of above evaluated at θ = θ 0 .<br />
Began with X ∼ MVN p (0,Σ) and showed<br />
X T Σ −1 X ∼ χ 2 p.<br />
Then started general theory of quadratic form:<br />
X t BX = (AZ) T B(AZ) = Z T QZ<br />
where Q = A T BA and AA T = Σ.<br />
Write Q = PΛP T where P is orthogonal (PP T = P T P = I) and Λ is<br />
diagonal. Columns of P are eigenvectors of Q and diagonal entries of<br />
Λ are eigenvalues.<br />
Richard Lockhart (<strong>Simon</strong> <strong>Fraser</strong> <strong>University</strong>) STAT 801=830 Statisti<strong>ca</strong>l Inference Fall 2012 2 / 3
Coverage in the text<br />
Method of moments in Chapter 9.<br />
See “course notes” on web pages <strong>13</strong>0-<strong>13</strong>1.<br />
Richard Lockhart (<strong>Simon</strong> <strong>Fraser</strong> <strong>University</strong>) STAT 801=830 Statisti<strong>ca</strong>l Inference Fall 2012 3 / 3