Poisson Regression Notes

# i i for a model with p n parameters induces i yi

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: here all λi s are different. ˆ The MLE is λi = Yi . The maximum value of the log-likelihood is yi log (yi ) − l (bmax ; Y ) = yi − i log (yi !) i i ˆ ˆ ˆ For a model with p < n parameters, β induces λi = Yi . The maximum log-likelihood value is ˆ yi log (yi ) − l (b; Y ) = i ˆ yi − i log (yi !) i UNM The deviance is, D = −2[l (b; Y ) − l (bmax ; Y )] or ˆ yi log (yi /yi ) − D=2 i ˆ (yi − yi ) i [oi log (oi /ei ) − (oi − ei )] =2 i =2 oi log (oi /ei ). i Since for most cases i oi = i ei (model with β0 ). The deviance residuals are deﬁned as di = sign(oi − ei ) [oi log (oi /ei ) − (oi − ei )] and so di2 . D= i UNM With a ﬁrst order Taylor ser...
View Full Document

## This note was uploaded on 01/27/2014 for the course STAT 574 taught by Professor Gabrielhuerta during the Fall '13 term at New Mexico.

Ask a homework question - tutors are online