Hi there,
I got a question that is both about stats and R. Imagine two alternative
stochastic processes:
a. Each subject has an exponential distribution of lifetime, with a parameter
lambda; If I simulate 100 such subjects and rank their lifetimes, they'd
look like this:
>plot(sort(rexp(100), decreasing=T))
b. The alternative process is slightly different. Imagine that the first subject
obeys the same rule as (a) and therefore would have an expected lifetime of
1/lambda. The second subject would have half of the lifetime of the first
subject (i.e., 1/(2*lambda)), the third would have one third of the lifetime as
the first subject (1/(2*lambda), and so on.
The distribution of ranked data of process (b) would therefore be more concave
than (a).
Now, here's my question. If I have a given dataset of subject lifetimes, how
can I specify and test these alternative processes in R?
Any help will be greatly appreciated.
Jonathan
[[alternative HTML version deleted]]