seqgen.missing: Generate random missing elements within a state sequence...

In TraMineRextras: TraMineR Extension

Generate random missing elements within a state sequence object

Description

The function assigns missing values (nr attribute of the object, which is "*" by default) to randomly selected positions in randomly selected cases.

Usage

seqgen.missing(seqdata, p.cases = 0.1, p.left = 0.2, p.gaps = 0, p.right = 0.3,
          mt.left="nr", mt.gaps="nr", mt.right="nr")

Arguments

seqdata	A state sequence object.
p.cases	Proportion of cases with missing values.
p.left	Proportion of left missing values.
p.gaps	Proportion of gap missing values.
p.right	Proportion of right missing values.
mt.left	Type of left missing. One of "nr" (non response state) or "void"
mt.gaps	Type of gap missing. One of "nr" (non response state) or "void"
mt.right	Type of right missing. One of "nr" (non response state) or "void"

Details

The aim of the function is essentially pedagogical. It may serve to illustrate how results of a sequential analysis may be affected by the presence of random missing states.

States in the sequences are randomly replaced with missing values. For each selected sequence, first, a random proportion between 0 and p.gaps of gaps are randomly inserted, then a random proportion between 0 and p.left of positions from the start of the sequence are set as missing, and finally a random proportion between 0 and p.right of positions from the end of the sequence are set as missing. Left missing values may possibly overlap gaps, and right missing values may overlap gaps and/or right missing values.

Value

The resulting state sequence object.

Warning

This function needs further testing.

Author(s)

Gilbert Ritschard

See Also

seqdef

Examples

## create the biofam.seq state sequence object from the biofam data.
data(biofam)
biofam.seq <- seqdef(biofam[1:100,10:25])

## Generate missing states within 60% of the sequences.
biofamm.seq <- seqgen.missing(biofam.seq, p.cases=.6,
      p.left=.4, p.gaps=.2, p.right=.5)

## compare the rendering of the sequences before and after
##   introducing missing states.
par(mfrow=c(2,2))
seqIplot(biofam.seq, sortv="from.end", with.legend=FALSE)
seqIplot(biofamm.seq, sortv="from.end", with.legend=FALSE)
seqdplot(biofam.seq, with.missing=TRUE, border=NA, with.legend=FALSE)
seqdplot(biofamm.seq, with.missing=TRUE, border=NA, with.legend=FALSE)
dev.off()

TraMineRextras documentation built on Sept. 11, 2024, 6:52 p.m.