Poisson Regression Proofs

Property 1: The maximum of the log-likelihood statistic for Poisson Regression occurs when the following k equations hold

Poisson regression maximize LL

Proof: As usual, our goal is to look at the xij and yi values as fixed and determine the unknown βj which maximize LL. We do this by setting the partial derivatives of LL with respect to the μi  to zero.

First, note that

Poisson regression mu derivative

ThusDerivative LL Poisson regression

Property 2: Let B = [bj] be the × 1 column vector of logistic regression coefficients, let Y = [yi] be the n × 1 column vector of observed outcomes of the dependent variable, let X be the × k design matrix (see Definition 3 of Least Squares Method for Multiple Regression), let P = [pi] be the n × 1 column vector of predicted values of success and V = [vij] be the n × n diagonal matrix where vij = zij on the main diagonal and zeros elsewhere. Then if B0 is an initial guess of B and for all m we define the following iteration

Logistic regression iteration Newton

then for m sufficiently large  B ≈ Bmand so Bm is a reasonable estimate of the coefficient vector.

Proof: The proof follows from the following observation

Second derivative Poisson regression

References

Hintze, J. L. (2007) Poisson regression. NCSS
https://www.ncss.com/wp-content/themes/ncss/pdf/Procedures/NCSS/Poisson_Regression.pdf

Nussbaum, E. M., Elsadat, S., Khago, A. H. (2007) Best practices in evaluating count data, Chapter 21: Poisson regression.
http://www.academia.edu/438746

Penn State (2017) Poisson regression. STAT 504: Analysis of discrete data.
https://online.stat.psu.edu/stat504/lesson/9

Leave a Comment