I'm trying to follow random survival forests. The ensemble mortality for $x_i$ is estimated by $ M_{e,i}=\sum_{j=1}^nH_e(T_j|x_i)$. Which means - sum over all death/censor times j, and for each such time get the averaged ensemble estimation for $x_i$ in time $j$.

I think that this means that for each tree in the ensemble, we detect the terminal node h in which $x_i$ sits, and sum its CHF $H_h$ over all possible points in time.

Under the conservation of events, this simply means sum the number of deaths in the terminal node of each tree in the ensemble! Is this correct? Is this is how its implemented in the R package? (can't read R for the life of me). If this is indeed so simple, why isn't it stated in the paper?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.