Nothing
#' @title Imputation under MCAR and MNAR hypothesis
#'
#' @description
#' this function performs missing values imputation under MCAR and MNAR hypothesis
#'
#' @param dataSet.mvs expression matrix containing abundances with MVs
#' (either peptides or proteins)
#' @param model.selector - binary vector; "1" indicates MCAR proteins
#' @param method.MAR - the method to be used for MAR missing values
#' - possible values: MLE (default), SVD, KNN
#' @param method.MNAR - the method to be used for MAR missing values
# - possible values: QRILC (default), MinDet, MinProb
#'
#' @return dataset containing complete abundances
#'
#' @export
#'
impute.MAR.MNAR = function(dataSet.mvs, model.selector,
method.MAR = "KNN", method.MNAR = "QRILC"){
# ___________________________________________________________________________________
# perform MAR imputation using the specified method
# -----------------------------------------------------------------------------------
switch(method.MAR,
MLE = {
dataSet.MCAR.imputed = impute.MAR(dataSet.mvs,
model.selector,
method = "MLE")
},
SVD = {
dataSet.MCAR.imputed = impute.MAR(dataSet.mvs,
model.selector,
method = "SVD")
},
KNN = {
dataSet.MCAR.imputed = impute.MAR(dataSet.mvs,
model.selector,
method = "KNN")
}
)
# ___________________________________________________________________________________
# perform MAR imputation using the specified method
# -----------------------------------------------------------------------------------
switch(method.MNAR,
QRILC = {
dataSet.complete.obj = impute.QRILC(dataSet.MCAR.imputed,tune.sigma = 0.3)
dataSet.complete = dataSet.complete.obj[[1]]
},
MinDet = {
dataSet.complete = impute.MinDet(dataSet.MCAR.imputed)
},
MinProb = {
dataSet.complete = impute.MinProb(dataSet.MCAR.imputed)
}
)
return(dataSet.complete)
}
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.