mt (version 2.0-1.20)

mv.util: Missing Value Utilities

Description

Functions to handle missing values of data set.

Usage

mv.stats(dat,grp=NULL,...) 
  
mv.fill(dat,method="mean",ze_ne = FALSE)
mv.zene(dat)

Value

mv.fill returns an imputed data frame.

mv.zene returns an NA-filled data frame.

mv.stats returns a list including the components:

• mv.overall: Overall missng value rate.

• mv.var: Missing value rate per variable (column).

• mv.grp: A matrix of missing value rate for different groups if argument grp is given.

• mv.grp.plot: An object of class trellis for plotting of mv.grp if argument grp is given.

Arguments

dat

A data frame or matrix of data set.

grp

A factor or vector of class.

method

Univariate imputation method for missing value. For details, see examples below.

ze_ne

A logical value indicating whether the zeros or negatives should be treated as missing values.

...

Additional parameters to mv.stats for plotting using lattice.

Author

Wanchang Lin

Examples

data(abr1)
dat <- abr1$pos[,1970:1980]
cls <- factor(abr1$fact$class)

## fill zeros with NAs
dat <- mv.zene(dat)

## missing values summary
mv <- mv.stats(dat, grp=cls) 
plot(mv$mv.grp.plot)

## fill NAs with mean
dat.mean <- mv.fill(dat,method="mean")

## fill NAs with median
dat.median <- mv.fill(dat,method="median")

## -----------------------------------------------------------------------
## fill NAs with user-defined methods: two examples given here.
## a.) Random imputation function:
rand <- function(x,...) sample(x[!is.na(x)], sum(is.na(x)), replace=TRUE)

## test this function:
(tmp <- dat[,1])        ## an vector with NAs
## get the randomised values for NAs
rand(tmp)

## fill NAs with method "rand"
dat.rand <- mv.fill(dat,method="rand")

## b.) "Low" imputation function:
"low" <- function(x, ...) {
  max(mean(x,...) - 3 * sd(x,...), min(x, ...)/2)
}
## fill NAs with method "low"
dat.low <- mv.fill(dat, method="low") 

## summary of imputed data set
df.summ(dat.mean)