b*******n 发帖数: 605 | 1 比如说我有一个categorical data, and the assumption is: cell counts have
Poisson distributions. 用的是log-linear model. Sometimes we can get formulas
of MLE for parameters in the log-linear model. But sometimes we can't. In SAS,
the MLE can be calculated numerically. (The formulas are to get parameter
estimates from the categorical cell counts, let's call them functions of cell
counts for parameters). Here is my question:
In Birch's Theorem, one of the conditions is: parameters should be able to be
written |
|