Displaying 1 result from an estimated 1 matches for "m1pred".
2010 Aug 30
1
'mgcv' package, problem with predicting binomial (logit) data
...#39;cr'), family=binomial(logit))
summary(test)
plot(test, shade=TRUE)
gam.check(test)
test.pred <- predict(test,newdata=data,se.fit=TRUE,type='response',
na.action=na.omit)
I1<-order(age)
plot(age[I1], test.pred$fit[I1],lty=1, type="l")
lines(age[I1],test.pred$fit[I1]+2*M1pred$se[I1],lty=2)
lines(age[I1],test.pred$fit[I1]-2*M1pred$se[I1],lty=2)
plot(age[I1], test.pred$fit[I1] , type="l")
I a second step, I want to calculate a similar model, but only for
respondents with a certain characteristic. For example, in this case, only
for male respondents.
I use a cod...