Prevalence division option

[fredjaya]

Not yet deployed as of writing this, but a9a5c6a introduces an option to divide prevalence and CI/CrI outputs by a provided value (2000 by default). Minimal testing completed, but some outstanding issues/to-dos:

• Does not multiply prevalence by default
• No sanity checks as of yet
• Rename columns when applied (e.g. Prevalence per N units)
• Does not update automatically

[AngusMcLure]

I had quick play with this on my machine. Sorry, there was a bit of a confusion and I perhaps should have picked up on this from our conversation. The prevalence value outputted by PoolPrev etc. needs to be multiplied by a value, not divided by it.

Multiplying by 100 would give the prevalence per 100 units (i.e. the prevalence as a percentage)

Multiplying by 2000 would give you the prevalence per 2000 units etc.

I think the confusion comes because we talk about prevalence per X.

[fredjaya]

Thanks for catching that, will push and deploy shortly.

Was getting it running locally straightforward?

[AngusMcLure]

Getting it running locally was not hard, but I had to reinstall all the packages that it depended on into the renv thing. This took a while. One suggestion to make this faster is to have a commented out line in global.R or app.R which installs all the packages that have a library() call somewhere in the app. That way you just run that line once on your local machine. Thoughts? There may be better ways to do this, but I don't really know how renv works

[fredjaya]

Will need to double check but I think the workflow should be something like:

renv::install()
renv::activate()

Then run app

[caitlinch]

I updated the prevalence multiplication code for both MLE and Bayesian.
When N != 1, any prevalence columns in the output data frame are multiplied by that N, and the prevalence columns are renamed to append "per N units".
When N=1, there is no change to the prevalence columns.

I implemented the boolean flag documented by Fred for when no multiplication is desired, which sets N=1.

The ICC columns are not affected by N and do not undergo multiplication.

Still no sanity checks - specifically, consider negative N values.