This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: # calculation of MLE for Gamma distribution # illinois 1960 rainfall data (see chapter 10) # This function computes the method of moments estimator and its approximate variance gamma.moment.est.2 = function (x) { note = "Fit a Gamma distribution by method of moments" mu.1 = mean(x) mu.2 = mean(x^2) var.x = mu.2mu.1^2 V.hat = var(cbind(x,x^2)) lam.hat = mu.1/(mu.2mu.1^2) alpha.hat = mu.1*lam.hat # estimate of variance for lam.hat based on delta method a2 = array( c(mu.1^2+mu.2, mu.1)/((mu.2mu.1^2)^2) ,c(2,1)) tau.lam = t(a2)%*%V.hat%*%a2 a1 = array( c( 2*mu.1*mu.2, mu.1^2)/((mu.2mu.1^2)^2) ,c(2,1)) tau.alpha = t(a1)%*%V.hat%*%a1 list(theta.hat=c(lam.hat,alpha.hat),mu.1=mu.1,mu.2=mu.2,V.hat=V.hat,tau1=tau.alpha,tau2=tau.lam) } # the MLE needs to be computed using an interative method, as there is no explicit solution to the MLE # it appears easier to use the minimizer optim, but two R functions to find the MLE are given here # this function gives the negative log likelihood in a form to be used in the nonlinear minimier nlm NLik = function(a,x){ lam = a[1] alpha = a[2] n = length(x) L = n*alpha*log(lam) + (alpha1)*sum(log(x))  lam*sum(x)  n*log(gamma(alpha)) L = (1)*L # .grad is a 1 by 2 matrix that is the derivatives of L .grad = array(0, c(1,2), list(NULL, c("lam", "alpha")))....
View
Full
Document
 Spring '11
 qqqq

Click to edit the document details