transitions: Transitions between elements of a sequence

In gobbios/cfp: Christof's functions

Transitions between elements of a sequence

Description

Transitions between elements of a sequence

Usage

transitions(S, strucmat = NULL, xstart = TRUE, xstop = FALSE, out = "ml")

Arguments

S	character, the sequence
strucmat	an optional matrix with column and row names (see details and examples)
xstart	logical, should start state be included
xstop	logical, should end state be included
out	character: either "ml" or "bayes"

Details

usage of strucmat might be helpful for at least tow reasons: 1) if there are constraints on the possible transitions. For example, a start cannot be followed by a stop (that wouldn't be a sequence). Such impossible transitions are indicated by 0 in this matrix. 2) if sequences are analysed that not all contain the same/complete set of elements. This might be useful if you want to generate matching outputs even though the actual observed sequences (and hence the result of the function) might differ (e.g. in the case of letter transitions in words)

References

Alger, S. J., Larget, B. R., & Riters, L. V. (2016). A novel statistical method for behaviour sequence analysis and its application to birdsong. Animal behaviour, 116, 181-193.

Examples

transitions("AAAAABBC", out="ml")
transitions("AAAAABBC", out="bayes")

smat <- matrix(1, ncol=3, nrow=4);
colnames(smat) <- c("A", "B", "C"); rownames(smat) <- c("start", "A", "B", "C")

transitions("AAAAABBC", out="ml", strucmat = smat)
transitions("AAAAABBC", out="bayes", strucmat = smat)

# errors:
# transitions("AAAAABBC", out="ml", strucmat = smat, xstop = TRUE)
# transitions("AAAAABBC", out="bayes", strucmat = smat, xstop = TRUE)

# add a stop column, but constrain that starts cannot be followed by stops
smat <- cbind(smat, 1); colnames(smat)[4] <- "stop"; smat[1, 4] <- 0
transitions("AAAAABBC", out="ml", strucmat = smat, xstop = TRUE)
transitions("AAAAABBC", out="bayes", strucmat = smat, xstop = TRUE)

gobbios/cfp documentation built on April 11, 2022, 2:22 a.m.