imputeForest: imputeForest

In dsnavega/imputeForest: Missing value imputation model using randomForest

Description

Missing Value Imputation through randomForest Proximity Matrix

Usage

imputeForest(x, y = NULL, method = "missForest", initial = "mode",
  mtry = ceiling(sqrt(ncol(x))), ntree = 300, iterations = 5,
  mcar = 0.1, crossvalidation = T, kfold = 5, verbose = F,
  seed = 1984, ...)

Arguments

x	a data.frame.
y	a numeric or factor vector.
method	type of procedure for imputation. Default is "proximity" for proximity based imputation. "missForest" triggers a hybrid technique using missForest and proximity based imputation.
initial	a character defining how initial missing values are imputed. Default is "mode" for mode imputation. "random" for random imputation from marginal distribution of each feature.
mtry	number of varibles to try at each split.
ntree	number of trees.
iterations	number of iterations used to update imputed values.
mcar	amount of missing values to induced completely at random. Used compute cross-validation imputation error.
crossvalidation	a logical indicating whether cross-validation should be made or not. Default is TRUE.
kfold	number of fold in cross-validation. Default is 5.
verbose	a logical indicating if output should be more detailed.
seed	a positive integer defining the seed for the random number generator.
...	additional arguments for randomForest

Value

an object of class imputeForest

Author(s)

David Navega

See Also

randomForest

Examples

x <- iris[, -5]
y <- iris$Species
x_na <- induce_missing(x)
print(imputeForest(x = x_na, y = y))

dsnavega/imputeForest documentation built on May 8, 2019, 2:43 p.m.