Changing sequence time granularity by aggregating positions
Changes time granularity of a state sequence object by aggregating successive positions into groups of a user-defined time length.
seqgranularity(seqdata, tspan = 3, method = "last")
seqdata |
A state sequence object. |
tspan |
Integer. Number of successive positions grouped together. |
method |
Character string. Aggregating method. One of |
Successive positions are aggregated by group of tspan states. The aggregated state is, depending of the method chosen, either the first ("first"), the first valid ("first.valid"), the last ("last"), the last valid ("last.valid"), or the most frequent ("mostfreq") state of the tspan long spell. The same applies to the last spell, even when it is shorter than tspan.
Methods ("first") and ("last") differ from ("first.valid") and ("last.valid") only when sequences contain missing values and/or have different lengths.
When there are (void or non void) missings, method "mostfreq" replaces each interval with the most frequent valid state on the interval or the missing state when there are no valid state.
End missings are set as void when there are voids in seqdata and as the nr attribute otherwise.
A stslist object: The compacted state sequence object.
Matthias Studer and Gilbert Ritschard
data(mvad) mvad <- mvad[1:100,] mvad.seq <- seqdef(mvad[,17:86], xtstep=12) mvadg.seq <- seqgranularity(mvad.seq, tspan=6, method="first") par(mfrow=c(2,1)) seqdplot(mvad.seq, with.legend=FALSE, border=NA) seqdplot(mvadg.seq, with.legend=FALSE)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.