f_train: Training xgboost models
In kristian-bak/kb.modelling: What the Package Does (One Line, Title Case)

Training xgboost models

f_train(
  df_train,
  target_name,
  var = NULL,
  nrounds = 32,
  max.depth = 4,
  eta = 1,
  min_child_weight = 50,
  early_stopping_rounds = 25,
  subsample = 0.9,
  colsample_bytree = 0.9,
  gamma = 1,
  ...
)

`df_train`	train data. Should be data.table object
`target_name`	target variable. Should be a string with a variable present in train data.
`var`	variables used to train the model. Should be a character vector. Default value is NULL, indicating that all variables available from f_indicators will be used.
`nrounds`	max number of iterations. Default is 32
`max.depth`	Max depth of the trees, i.e. the number of splits made.
`eta`	step size of each boosting step. Large eta may lead to unstable results. Default is 1.
`min_child_weight`	minimum counts in a child. The algorithm stops if splitting leads to leaf node with fewer than min_child_weight instances.
`early_stopping_rounds`	stopping criteria. If performance is not improved after k(= early_stopping_rounds) iterations, the algorithm stops.
`subsample`	train proportion for each fold in cross validation. Should be a number between 0 and 1.
`colsample_bytree`	subsample proportion for each tree. Should be a number between 0 and 1.
`gamma`	minimum loss reduction required to make a further partition on a leaf node of the tree. The larger, the more conservative the algorithm will be.
`...`	additional parameters passed to xgboost.