rings<-challenger$V1
temp<-challenger$V2
plot(temp, rings)
#If you look at only failures (ignore Y=0) then you do not see any pattern. 
# But the case of Y=0 is informative. You clearly see that Y=0 generally for higher values of X.
# We don't knw how to model this. Redife Y as indicator there was at least once O ring damage.
rings<-(rings>=1)
plot(temp, rings, ylab="Failure Probability")
m<-glm(formula=rings~temp, family=binomial)
summary(m)
exp(m$coefficients[2])
o<-order(temp)
x<-sort(temp)
y<-predict(m,type="response")[o]
lines(x,y)