BS2 Statistical Inference - Lecture 9
The method of scoring. The EM algorithm
This lecture describes the method of scoring for calculating the MLE. Fisher's method of scoring replaces the observed second derivative of the log-likelihood function with its expectation, the Fisher information. This may often be simpler and less unstable than the Newton-Raphson method.
The important EM algorithm is described, with examples to come in the next lecture.
Overheads
Links
Last updated: Friday, 12 November 2004 11:02. Steffen L. Lauritzen