[R-package] update parameter 'verbosity' based on keyword arg 'verbose'

[jameslamb]

Related to #4667.
Related to #4862.

While investigating #4667, I noticed a small inconsistency in lgb.train() and lgb.cv(). Those functions have a keyword argument verbose and update the parameter verbose in parameters passed down to LGBM_BoosterCreate.

LightGBM/R-package/R/lgb.cv.R

Line 105 in 7b10baf

params$verbose <- verbose

LightGBM/R-package/R/lgb.train.R

Line 77 in 7b10baf

params$verbose <- verbose

As a result, if you pass different values for verbosity (in params) and verbose (via keyword arguments), then the Booster's parameters will contain both verbosity and verbose.

Consider this example:

library(lightgbm)
data("agaricus.train")
dtrain <- lightgbm::lgb.Dataset(
    data = agaricus.train$data
    , label = agaricus.train$label
)
bst <- lightgbm::lgb.train(
    params = list(
        num_iterations = 2L
        , verbosity = 1L
    )
    , data = dtrain
    , obj = "binary"
    , verbose = -1L
)
bst$params

$verbosity
[1] 1

$verbose
[1] -1

Note that verbosity is the main parameter recognized by LightGBM (https://lightgbm.readthedocs.io/en/latest/Parameters.html#verbosity) and verbose is an alias for it.

This PR proposes two changes:

1. Update params$verbosity (the main parameter name, instead of an alias) based on the value of keyword argument verbose
2. Use internal function lgb.check.wrapper_param() to ensure consistent behavior when processing keyword arguments.

[jameslamb]

these uses of invisible(capture.output(...)) suppress log messages from being printed, per #4862

[StrikerRUS]

Nice improvement!

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.