This PR adds S3 methods for class pioneer_bootstrap. It also adds support for NULL value for the ids argument in compute_dea(). In addition to this, this PR does the following:
Rework some of the logic of the bootstrap functions in order to make the model object available when the S3 object is constructed
Update the server() function to work with the new bootstrap functions
Add support for NULL value for the ids argument in get_peers()
Optimise performance of get_peers() (about 3.5x faster)
The most "controversial" choice in this PR, is the support for excluding the need for the id argument in compute_dea(). This is done by setting id to NULL if the argument is missing. Without this, the function will fail if the user does not specify id, ie. compute_dea(data, input = c('col1', 'col2'), output = c('col3', 'col4')). However, this opens up the possibility of errors if the user relies on positional arguments, but skips id, ie. compute_dea(data, c('col1', 'col2'), c('col3', 'col4'), 'vrs')
This PR adds S3 methods for class
pioneer_bootstrap
. It also adds support forNULL
value for theids
argument incompute_dea()
. In addition to this, this PR does the following:server()
function to work with the new bootstrap functionsNULL
value for theids
argument inget_peers()
get_peers()
(about 3.5x faster)The most "controversial" choice in this PR, is the support for excluding the need for the
id
argument incompute_dea()
. This is done by settingid
toNULL
if the argument is missing. Without this, the function will fail if the user does not specifyid
, ie.compute_dea(data, input = c('col1', 'col2'), output = c('col3', 'col4'))
. However, this opens up the possibility of errors if the user relies on positional arguments, but skipsid
, ie.compute_dea(data, c('col1', 'col2'), c('col3', 'col4'), 'vrs')