R中的通用马尔可夫链

Question

I need construct a generic markov chain from the following data for the weather patterns in newyork. There are three regions- manhattan, bronx and queens- I need to have a overall markov chain graph that describes the average transition probabilities for the whole region.

state<- c('rainy', 'sunny', 'rainy', 'sunny', 'cold' ,'rainy', 'cold')
region<- c('manhttan', 'manhattan', 'bronx', 'bronx', 'queens', 'queens', 'queens')
df<- data.frame(region, state)

 library(markovchain)
 mcFit <- markovchainFit(data=df$state)
 print(mcFit)
 plot(mcFit$estimate, edge.arrow.size=.89)

any ideas?

Answer 1

The following gives a list of transition matrices by region:

M <- tapply(state, region, function(s) 
  markovchainFit(data = s)$estimate@transitionMatrix)

Hence, the average one then is

Reduce("+", M) / length(M)

In this case it doesn't really work though, because there is too little data to estimate a decent matrix for each region.