#learning GLMs
#tutorial from http://www.statmethods.net/advstats/glm.html
#survival analysis
#load library(lung) - lung cancer patient survival

#create a surv object
survobj<-with(lung,Surv(time,status))
#Surv creates a survival object as a response variable in a model formula; time = for right censored data; status = ending time of interval

#plot survival distribution of the total sample
#Kaplan-Meier estimator
fit0<-survfit(survobj~1, data=lung)
#survfit creates survival curves from a formula or model
summary(fit0)
plot(fit0,xlab='survival time in days', ylab='% surviving', yscale=100, main='survival distribution (overall)')
#think this plots survival by time with upper and lower 95% CI

#compare survival dist of men and women
fit1<-survfit(survobj~sex, data=lung)
plot(fit1, xlab='survival time in days', ylab='% surviving', yscale=100, col=c('red', 'blue'), main='survival distributions by gender')
legend('topright', title='gender', c('male', 'female'), fill=c('red', 'blue'))

#test for differences between male and female
#survival curves (logrank test)
survdiff(survobj~sex,data=lung)
#does Chi square

#predict male survival from age and medical scores
MaleMod<-coxph(survobj~age+ph.ecog+ph.karno+pat.karno,data=lung, subset=sex==1)