xg_train: Calibrate Model

In ArnaudBu/ezXg: Easy xbgoost wrapper

Description

xg_train trains an xgboost model in order on the data structure generated by the xg_load_data function.

Usage

xg_train(data, eta = 0.3, gamma = 0, max_depth = 6,
  colsample_bytree = 1, min_child_weight = 1, nrounds = 100,
  nthread = 2, verbose = 1, cv = 1, seed = 1, objective = "auto")

Arguments

data	Object. A data structure created by the call of the xg_load_data function.
eta	Numeric. Eta parameter for xgboost calibration. See xgb.train for more details.
gamma	Numeric. Gamma parameter for xgboost calibration. See xgb.train for more details.
max_depth	Numeric. Max_depth parameter for xgboost calibration. See xgb.train for more details.
colsample_bytree	Numeric. Colsample_bytree parameter for xgboost calibration. See xgb.train for more details.
min_child_weight	Numeric. Min_child_weight parameter for xgboost calibration. See xgb.train for more details.
nrounds	Numeric. Nrounds parameter for xgboost calibration. See xgb.train for more details.
nthread	Numeric. Nthread parameter for xgboost calibration. See xgb.train for more details.
verbose	Numeric. Verbose parameter for xgboost calibration. See xgb.train for more details.
cv	Numeric. Number of folds in cross validation. If this parameter is set to 1, this means that cross-validation will not be performed.
seed	Numeric. Seed for computation reproducability.
objective	Character. Objective function for the optimization. . Eta parameter for xgboost calibration. See xgb.train for more details. Can be set to auto in order to let the function choose the better model regarding the output variable.

Value

A trained model with following values:

• model: calibrated model as returned by the xgb.train function.

• ntree: optimal number of tree according to the test set.

• formula: the formula used for constructing the model matrix and that is applied when running the model.

• template: an empty data.table that has saved all the input values and that is used to appropriately format data when using the prediction function.

• labels: The possible labels for prediction when performing a classification task with xgboost.

• na.handle: passed to reapply to prediction

In case the parameter cv is set to anithing but 1, the function returns a 1 line data.table with the average error on the cross-validation.

Examples

d <- xg_load_data(system.file("extdata", "titanic.csv", package = "ezXg"),
               inputs = c("Pclass", "Sex", "Age", "SibSp",
                          "Parch", "Fare", "Embarked"),
               output = "Survived",
               train.size = 0.8)
md <- xg_train(d)

ArnaudBu/ezXg documentation built on Oct. 30, 2019, 4:59 a.m.