RMark (version 3.0.0)

MStruncate: Truncate capture histories for multi-state models

Description

Decompose full capture history to releases followed by k recapture occasions. If a recapture occasion occurs before k occasions, the capture history is finished at the first recapture and right padded with "." which effectively acts like a loss on capture. The recapture is then a new release and new capture history. If there are no recaptures within k occasions, it has a release followed by k 0's. If the release is such that adding k occasions is greater than the length of the original capture history, then the new history is left padded with 0's. Capture histories that end with a capture on the last occasion do not generate a new capture history because there are no possible recaptures and thus contain no information in a CJS format MS model. All freq and covariates are copied with newly generated truncated capture histories.

Usage

MStruncate(data, k = 5)

Value

dataframe with field ch and freq (default to 1) and any covariates included in argument data; it also contains a factor variable release which is the first occasion which should be used as a group variable so the begin.time can be set for each release cohort to maintain the original times, as shown in the example.

Arguments

data

dataframe containing at least one character field named ch; can also contain frequency in numeric field freq and any other covariates.

k

number of recapture occasions after release; new capture histories are of length k+1

Examples

# \donttest{
data(mstrata)
df=MStruncate(mstrata,k=2)
dp=process.data(df,model="Multistrata",groups=c("release"),begin.time=1:max(as.numeric(df$release)))
ddl=make.design.data(dp)
table(ddl$S$release,ddl$S$time)
table(ddl$p$release,ddl$p$time)
# }