In: Statistics and Probability
Let X~Geometric(p), with parameter p unknown, 0<p<1.
a) Find I(p), the Fisher Information in X about p.
b) Suppose that pˆ is some unbiased estimator of p. Determine the Cramér-Rao Lower Bound for Var p[ ]ˆ based on one observation from this distribution.
c) Show that p= I {1}(X) is an unbiased estimator of p. Does its variance achieve the Cramer-Rao Lower Bound?
Answer:-
Given That:-
Let X~Geometric(p), with parameter p unknown, 0<p<1.
a) Find I(p), the Fisher Information in X about p.
Fisher information in X about p is
Now,
ln f(x; p) = ln p + (x - 1) ln (1 - p) fro x = 1, 2, -------
b) Suppose that is some unbiased estimator of p. Determine the Cramér-Rao Lower Bound for Var p[ ]ˆ based on one observation from this distribution.
Based on one observation X from this distribution CRLB fro var (); being an Unbiased estimator of p; is
c) Show that p= I {1}(X) is an unbiased estimator of p. Does its variance achieve the Cramer-Rao Lower Bound?
Hence,
is an unbiased estimator of p.
= p[X = 1] - p2 [X = 1] = p - p2 = p(1 - p)
Which is greater than p2(1 - p) = CRLB
Thus variance of does not achieve CRLB.