gl<-glm(SF~s,family=binomial(link='logit'))

advertisement
link = function of the mean response
gl<-glm(SF~s,family=binomial(link='logit'))
Explanatory variable
Response variable; for binomial link can be a
two-column matrix with success/failure counts
Family = distribution of response
“response” = compute the response variable
“link” = compute the link function values
pl<-predict(gl,data.frame(s=x),type='response')
“x” is the grid on which the
predicted values will be calculated
Result of the glm()
“s” is the name that we used
in our call to glm()
Deviance residuals show how much each of the observations
contributes to the total deviance
ML estimation of the model coefficients; standard error
(standard deviation of the estimator); corresponding z-value
under H0 that parameter is 0; and Prob(|z|>|parameter|),
where z~N(0,1)
Null deviance: Deviance of the NULL model, which assumes
that all parameters (except intercept) equal to 0.
Residual deviance: Deviance + a constant chosen in such a way
that the saturated model’s deviance is 0.
Akaike Information Criterion
AIC = -2 Log-Likelihood +2 # of parameters
Download