ensembleImpute: Computes the mean, median, or mode among a list of impute.yai...

In yaImpute: Nearest Neighbor Observation Imputation and Evaluation Tools

Computes the mean, median, or mode among a list of impute.yai objects

Description

Several objects of class impute.yai or yai are combined by computing the mean, median, or mode of separate, individual imputations. The intention is that the members of the first argument would be formed by running yai several times with bootstrap=TRUE or by varying other options.

Usage

ensembleImpute(imputes, method="mean",...)

Arguments

imputes	a list of objects class impute.yai or yai. Function impute.yai is called for list members where the class is yai.
method	when "mean", the continuous variables are averaged using mean, otherwise the median is used. Mode is always used for character data (generally the case for factors).
...	passed to impute.yai.

Value

An object of class c("impute.yai","data.frame"), see impute.yai. The attributes of the data.frame include the following:

• sd A data.frame of standard deviations for continuous variables if there are any. The columns are not reported if the standard deviation is zero for all observations which is typically true of "observed" values.

• N the number of replications used to compute the corresponding data; reported only if the number differs from the total number of replications. This will be the case when bootstrap, sampleVar, or both are used in yai.

• methods the method used for each variable.

Author(s)

Nicholas L. Crookston [e-mail]
John Coulston [e-mail]

See Also

yai buildConsensus impute.yai

Examples

require (yaImpute)
data(iris)

set.seed(123) 

# form some test data, y's are defined only for reference
# observations.
refs=sample(rownames(iris),50)
x <- iris[,1:2]      # Sepal.Length Sepal.Width
y <- iris[refs,3:4]  # Petal.Length Petal.Width

reps <- replicate(10, yai(x=x,y=y,method="msn",bootstrap=TRUE,k=2),
                  simplify=FALSE)

ensembleImpute(reps,ancillaryData=iris)

yaImpute documentation built on Nov. 4, 2022, 1:06 a.m.