Skip to contents

Create cross-validation training data sets.

create_folds.Rd

Creates a list of training datasets each of which has certain entries of remove_cols replaced by NA according to the type of cross-validation. If type = "LOO" one entry of each training dataset is replaced by NA, if type = "SLOO" one entry and the entries of the neighbours of the area it corresponds to are replaced by NA.

Usage

create_folds(sf, remove_cols = c("y"), type = "LOO")

Arguments

sf

A simple features object with some geometry.

remove_cols

A vector of named columns which are to have entries replaced by NA in the training data sets. Defaults to c("y").

type

One of "LOO" or "SLOO".

Value

A list of nrow(sf) training set lists. Each training set list contains:

  • data The training data set with left-out entries.

  • held_out The indices of all held-out regions.

  • predict_on The indices of regions to be predicted upon.

Examples

create_folds(mw, remove_cols = c("y", "est"))
#> Error in remove_cols %in% names(sf): object 'mw' not found