Other cause mortality

[RobertASmith]

Hi @n8thangreen,

Was just going through this paper. Very useful addition to the literature.

Had two complaints (because I'm a miserable sod).

1. The cycles going across columns. Will ignore this.
2. The lookup for other cause mortality occurring within the function creating transition probabilities.

This pull request has a suggestion for 2 - creating the age specific other cause mortality in a vector outside the function. This should speed things up a bit.

Kind regards,

Rob

[n8thangreen]

Hi @RobertASmith
Thanks for this.

1. I remember Paul didn't like this either. I think we decided to structure like this so that the print statement looked the nicest.
2. So this is precomputing the table and passing this in? Run time was not a problem for our model and hiding the lookup in the function means we pass one less thing around. I agree yours is more generalisable and no doubt faster. If say there were different life table for different countries I can see this advantage. We could also partially execute the trans prob function for a given lookup e.g. purrr::partial().

[RobertASmith]

The pull request is basically an excuse to say "nice paper".

Don't worry about (1), I won't tell Paul :)

I saw you are working on a decision tree package, quite interested to see where that goes, TreeAge is unnecessarily expensive.