crossv_df: Generate cross-validation test-training sets manually
In jrnold/resamplr: Resampling methods

Description Usage Arguments Methods (by class) Examples

Generate cross-validation test-training sets from manually specified indices of observations in the test and and training sets.

crossv_df(data, ...)

## S3 method for class 'data.frame'
crossv_df(data, train = NULL, test = NULL, ...)

## S3 method for class 'grouped_df'
crossv_df(data, train = NULL, test = NULL, ...)

`data`	A data frame
`...`	Arguments passed to methods
`train`	A list of integer vectors or an integer vector, with the indices of the training set. If `train` is `NULL`, then the training set is the complement of the test set.
`test`	A list of integer vectors or an integer vector, with the indices of the test set. If `test` is `NULL`, then the test set is the complement of the training set.

data.frame: Partition a data frame into test and training sets by row.
grouped_df: Partition a grouped data frame into test and training sets by group.

library("dplyr")

# test-training pairs can be specified by test sets
crossv_df(mtcars, train = list(1:5, 6:10))
# or by training sets
crossv_df(mtcars, train = list(1:20, 10:32))
# or by both (and they don't need to partition the observations)
crossv_df(mtcars,
          train = list(1:20, 10:32),
          test = list(21:32, 1:9))

# with grouped data frames, the indices refer to groups
crossv_df(group_by(mtcars, mpg), train = list(1L))$train