309 fixed power prior

[gravesti]

No description provided.

[mattsecrest]

@gravesti looks good. Was a bit surprised to see a vector of weights, as the fixed power prior is typically one weight for all the external data, right? (And I see that the test cases are doing this already). I guess there is no downside to doing it this way because it is more flexible and we can use it for eg IPTW weights.

Seeing now that the stan templates is annoying, but still I bet this is the best approach.

[gravesti]

Was a bit surprised to see a vector of weights, as the fixed power prior is typically one weight for all the external data, right? (And I see that the test cases are doing this already). I guess there is no downside to doing it this way because it is more flexible and we can use it for eg IPTW weights.

Yeah, we could go for specifying a number directly: borrowing_fixed_power_prior("ext", power = 0.7)
and then set that for the external subjects and the rest to 1.

[gravesti]

@mattsecrest I've changed to a single power_par value specified directly. Let me know what you think.

[mattsecrest]

@gravesti looks good! I like the templates in inst/stan more and more. There's one line that might be too clever?

power <- 1 - (1 - borrowing@power_par) * data_matrix[, borrowing@ext_flag_col]

Maybe this?

power <- rep(1, nrow(data_matrix))
power[data_matrix[, borrowing@ext_flag_col] == 1] <- borrowing@power_par

Before we submit:

• update vignettes
• CI/CD pass

[mattsecrest]

Addresses #309