zandergordan / mripr Goto Github PK

Currently, processing the raw data, and then performing calculations/estimations, is bundled together in a single function. This is not desirable behavior and needs to be refactored into 2 separate functions.

Currently, the API for the catch calculations leaves a lot to be desired. Things that could be improved:

• The first argument should be an R data.frame created by the data loader function, not a path to a data directory.
• the behavior of the domain argument should be adjusted to match the defaults of the web query. In particular, annual, US by sub-region, all modes combined and all areas combined should be assumed unless otherwise specified. The current default that common = NULL implies all species is good, and in fact is probably an improvement of the catch time series web tool.
• the Domain column in the returned data.frame is not useful. The fields used to specify the domain should be included separately in the returned object, as is standard practice in dplyr::group_by() aggregations.

.Rbuildignore needs to be defined such that extraneous files in the current repo will not interfere with R's package-building utilities.

A good candidate would be MRIPr, pronounced "m-ripper".

programs/ needs to be renamed R/ to comply with R package standards.

We need a function that downloads the MRIP data, processes it, and saves it locally.

This should be a brief wrapper around something like:

LoadRDS()

The purpose of which is to check some cached information about when the data in the .rds was downloaded, compare that to what is currently on the MRIP FTP site (if possible, ie. If there is internet connection), and download updated versions of the data if available.

Currently the raw csv data is stored in inst/extdata, when we should be storing the processed data in data/

Each function should have an @example tag in the roxygen2 section