train_lightgbm: train_lightgbm
In gm209/gmtools: A collection of generic ML functions for Insurance Risk Modelling

A wrapper function that automates the lightGBM training process

1
2
3

train_lightgbm(dtrain, dvalid = NULL, x, y, w = NULL, LgbParams,
  nrounds = 5000, stratified = FALSE, early_stopping_rounds = 5,
  nfold = 5, folds = NULL, verbose = 1, seed = 1921, ...)

`dtrain`	A data.frame that contains the training data
`dvalid`	A data.frame that contains the validation data. If NULL then the function uses cross validation. Defaults to NULL.
`x`	A list of column names that identify the explanatory features
`y`	A column name that identifies that target variable
`w`	A column name that contains the name of the weight column. Defaults to NULL
`LgbParams`	A named list containing the lightgbm Learning Parameters
`nrounds`	The maximum number of iterations
`early_stopping_rounds`	If performance doesnt improve for this many rounds then stop training
`nfold`	How many folds to use if doing cross validation
`folds`	A list of pre defined fold indicies (test indicies) - see main lightgbm docs for more details.
`verbose`	Print run time messages
`seed`	An integer which will be used as the random seed. Defaults to 1921
`...`	Additional Arguments to be passed to lgb.cv and/or lgb.train