RT Attrition Prediction problem
Given time interval: 0, … T, … T+D
Given customer c, with final state A(c)=0 or 1 at time T+D (e.g. 1 if customer attrites, 0 if not)
At any time 0 < t < T model M should be able to compute a probability M(c,t) of A(c)=1