calibrate: Calibrate parameters

In mcubillos3/biokNN: Bi-Objective k-Nearest Neighbors Imputation for Multilevel Data

Description

This function returns a vector with the two parameters requiered by the biokNN method where the first value is the weighting parameter and the second the number of neighbors

Usage

calibrate(
  data,
  className,
  varNames,
  prop_valid = 0.1,
  nIter = 10,
  distance = "gower",
  alpha_space = NULL,
  k_space = NULL,
  print = FALSE
)

Arguments

data	A dataframe with missing values
className	name of the variable that contains the classes
varNames	vector with the names of the variables to be imputed
prop_valid	proportion of missing values
nIter	number of iterations, default = 10
distance	distance function used to get the k-nearest neighbors
alpha_space	vector with the calibration values to test for the weight parameter
k_space	vector with the calibration values to test for the number of neighbors
print	option to print the RMSE values of the parameters used for calibration (print = TRUE).

Value

A dataframe with the imputed data

Examples

data(data_example)
calibrate(data_example,
          className = "class",
          varNames = c("y"),
          prop_valid = 0.3,
          alpha_space = c(0.5, 0.7, 0.9),
          k_space = c(10, 15),
          print = TRUE)

mcubillos3/biokNN documentation built on Dec. 21, 2021, 4:50 p.m.