Iscores (version 1.1.0)

Iscores: Iscores: compute the imputation KL-based scoring rules

Description

Iscores: compute the imputation KL-based scoring rules

Usage

Iscores(
  imputations,
  methods,
  X.NA,
  m = length(imputations[[1]]),
  num.proj = 100,
  num.trees.per.proj = 5,
  min.node.size = 10,
  n.cores = 1,
  projection.function = NULL,
  rescale = TRUE
)

Arguments

imputations

a list of list of imputations matrices containing no missing values of the same size as X.NA

methods

a vector of characters indicating which methods are considered for imputations. It should have the same length as the list imputations.

X.NA

a matrix containing missing values, the data to impute.

m

the number of multiple imputation to consider, defaulting to the number of provided multiple imputations.

num.proj

an integer specifying the number of projections to consider for the score.

num.trees.per.proj

an integer, the number of trees per projection.

min.node.size

the minimum number of nodes in a tree.

n.cores

an integer, the number of cores to use.

projection.function

a function providing the user-specific projections.

rescale,

a boolean, TRUE if the scores should be rescaled such that the max score is 0.

Value

a vector made of the scores for each imputation method.

Examples

# NOT RUN {
n <- 100
X <- cbind(rnorm(n),rnorm(n))
X.NA <- X
X.NA[,1] <- ifelse(stats::runif(n)<=0.2, NA, X[,1])

imputations <- list()

imputations[[1]] <- lapply(1:5, function(i) {
 X.loc <- X.NA
 X.loc[is.na(X.NA[,1]),1] <- mean(X.NA[,1],na.rm=TRUE)
 return(X.loc)
})

imputations[[2]] <- lapply(1:5, function(i) {
 X.loc <- X.NA
 X.loc[is.na(X.NA[,1]),1] <- sample(X.NA[!is.na(X.NA[,1]),1],
 size = sum(is.na(X.NA[,1])), replace = TRUE)
 return(X.loc)
})

methods <- c("mean","sample")

Iscores(imputations,
methods,
X.NA,
num.proj=5
)

# }