0 évaluation0% ont trouvé ce document utile (0 vote)
15 vues1 page
(1) The document discusses maximum likelihood estimation for several probability distributions. For an exponential distribution with parameter θ, the MLE of θ is consistent but not unbiased based on a sample of n observations. The MLE of λ is not necessarily unbiased.
(2) For a uniform distribution on (0,θ), the MLE θˆMLE = X(n) , the maximum observation, is shown to be consistent. Its distribution and mean are derived. An unbiased estimator θˆU is obtained by removing bias.
(3) For a binomial distribution with parameter p, the MLE pˆ = X/n, where X is the number of successes, is shown to be consistent
(1) The document discusses maximum likelihood estimation for several probability distributions. For an exponential distribution with parameter θ, the MLE of θ is consistent but not unbiased based on a sample of n observations. The MLE of λ is not necessarily unbiased.
(2) For a uniform distribution on (0,θ), the MLE θˆMLE = X(n) , the maximum observation, is shown to be consistent. Its distribution and mean are derived. An unbiased estimator θˆU is obtained by removing bias.
(3) For a binomial distribution with parameter p, the MLE pˆ = X/n, where X is the number of successes, is shown to be consistent
(1) The document discusses maximum likelihood estimation for several probability distributions. For an exponential distribution with parameter θ, the MLE of θ is consistent but not unbiased based on a sample of n observations. The MLE of λ is not necessarily unbiased.
(2) For a uniform distribution on (0,θ), the MLE θˆMLE = X(n) , the maximum observation, is shown to be consistent. Its distribution and mean are derived. An unbiased estimator θˆU is obtained by removing bias.
(3) For a binomial distribution with parameter p, the MLE pˆ = X/n, where X is the number of successes, is shown to be consistent
1. Let X be distributed with density function fX (x; ) = 1 ex/ , for x > 0.
(Remark: This is the usual exponential distribution parameterized in terms of the mean ; an alternative parameterization is in terms of = 1/, in which case the mean is 1/ and the variance 1/2 ) (a) Obtain the MLE of based on a sample of n independent observations. (b) Is it: i) Consistent? ii) Unbiased? (c) Is the MLE estimator of unbiased? 2. Consider the MLE estimator of in a U (0, ) distribution, based on n independent observations X1 , . . . , Xn . We already know know M LE = X(n) = max{X1 , . . . , Xn }. (a) Show that it is consistent. (Hint: You can follow a direct approach here. Fix a neighborhood of of arbitrary width and show that for sufficiently large n, the probability of being there is as close to 1 as desired.) (b) Find the distribution, then the density of M LE = X(n) . (Hint: F (x) = P (M LE x) = P (n Xi x).) i=1
M LE
(c) Find the mean of M LE . Remove the bias to obtain an unbiased
estimator U . (d) Find the variance of U . (Hint: Make use of the fact that 2 = 2 m2 ; you have computed the mean m in the previous step, so you only need 2 .) (e) What is the convergence rate to zero of the variance you have found? 3. When we take a sample of n independent observations X1 , . . . , Xn from a b(p) distribution, we have found that the MLE is X, the number of ones divided by the number of throws. Show that it is is consistent.