reprocows: Reproductive data from seven dairy cows

In jaredo/mhsmm: Inference for Hidden Markov and Semi-Markov Models

Description

This data set contains hourly observations on progesterone and an activity index at hourly intervals since calving on seven dairy cows.

Usage

reprocows

Format

reprocows is a data frame containing 13040 rows. id is the cow ID, progesterone is a measurement of the hormone in ng/L taken from a milk sample, activity is a relative measure of activity calculated from a pedometer.

There are a large number of missing values as progesterone is measured only at milking time (and at a farm manager's discretion). Missing values in activity occur due to hardware problems can occur with pedometers.

Source

Danish Cattle Research Centre

References

Peters, A. and Ball, P. (1995), "Reproduction in Cattle," 2nd ed.

Examples

data(reprocows)
data(reproai)
data(reproppa)
tm = 1600

J <- 3
init <- c(1,0,0)
trans <- matrix(c(0,0,0,1,0,1,0,1,0),nrow=J)
emis <- list(mu=c(0,2.5,0),sigma=c(1,1,1))

N <- as.numeric(table(reprocows$id))
train <- list(x=reprocows$activity,N=N)
class(train) <- "hsmm.data"
tmp <- gammafit(reproppa * 24)
M <- max(N)

d <- cbind(dgamma(1:M,shape=tmp$shape,scale=tmp$scale),
 # ppa sojourn directly estimated from ppa data set
dunif(1:M,4,30),
 # oestrus between 4 and 30 hours
dunif(1:M,15*24,40*24))
 #not-oestrus between 15 and 40 days

startval <- hsmmspec(init,trans,parms.emission=emis,list(d=d,type='gamma'),
  dens.emission=dnorm.hsmm)
#not run (takes some time)
#h.activity <- hsmmfit(train,startval,mstep=mstep.norm,maxit=10,M=M,lock.transition=TRUE)
  

jaredo/mhsmm documentation built on Dec. 6, 2019, 11:07 a.m.