10.3 Meaningful probabilities
It is meaningful to estimate (and calculate)
\[\begin{equation*}
P_k(t) = \int_0^t S(s) \alpha_k(s) ds,\quad t > 0,\quad k = 1, 2, 3,
\end{equation*}\]
the probability to die from cause \(k\) before time \(t\).
Note that
\[\begin{equation*}
S(t) + P_1(t) + P_2(t) + P_3(t) = 1 \texttt{ for all $t > 0$}.
\end{equation*}\]
Now, estimation is straightforward with the following estimators:
\[\begin{equation*}
\hat{P}_k(t) = \sum_{i:t_i \le t} \hat{S}(t_i-)\frac{d_i^{(k)}}{n_i}
\end{equation*}\]
See the function comp at the end of this chapter for how to write code for
this estimation.