xgb.Booster objects are created from the xgboost package,
which provides efficient and scalable implementations of gradient
boosted decision trees. Given the reliance of post processing
functions on the model object, like xgb.Booster.complete
,
on the first class listed, the butcher_xgb.Booster
class is
not appended.
Examples
library(xgboost)
#>
#> Attaching package: ‘xgboost’
#> The following object is masked from ‘package:dplyr’:
#>
#> slice
library(parsnip)
data(agaricus.train)
bst <- xgboost(data = agaricus.train$data,
label = agaricus.train$label,
eta = 1,
nthread = 2,
nrounds = 2,
eval_metric = "logloss",
objective = "binary:logistic",
verbose = 0)
out <- butcher(bst, verbose = TRUE)
#> ✔ Memory released: 31.12 kB
#> ✖ Disabled: `print()`, `summary()`, and `xgb.Booster.complete()`
#> ✖ Could not add <butchered> class
# Another xgboost model
fit <- boost_tree(mode = "classification", trees = 20) %>%
set_engine("xgboost", eval_metric = "mlogloss") %>%
fit(Species ~ ., data = iris)
out <- butcher(fit, verbose = TRUE)
#> ✖ The butchered object is 1.14 kB larger than the original. Do not butcher.