R help - Mar 2012 - Updating a Markov Chain

Hello,

my question is if anyone has any good ideas how to create a Markov Chain
from ordered data. So, I have some sort of time series, and if value1
happens as time1 and value2 happens at time2 I record this as an update to
the probability transition matrix. The problem is that I cannot predefine
the size of the matrix (as I don't know how many states(values) I will have
in the end) and don't really know how to update the prob distribution in the
rows. 
If anyone has any thoughts, i would be grateful for sharing.
Stivi

--
View this message in context:
http://r.789695.n4.nabble.com/Updating-a-Markov-Chain-tp4504156p4504156.html
Sent from the R help mailing list archive at Nabble.com.

On Sun, Mar 25, 2012 at 3:50 PM, stivi <muhamed19 at wp.pl>
wrote:
> Hello,
>
> my question is if anyone has any good ideas how to create a Markov Chain
> from ordered data. So, I have some sort of time series, and if value1
> happens as time1 and value2 happens at time2 I record this as an update to
> the probability transition matrix. The problem is that I cannot predefine
> the size of the matrix (as I don't know how many states(values) I will
have
> in the end)
You could create a iter x n matrix of NA's and fill it up with a loop.
Or an empty list
 ll <- list()
 for(i in 1:3) ll[[i]] <- sample(10,i)
 ll
Or, not very efficient, but you can also avoid predefining with something like
trans.prob <- .5
set.seed(1)
for(i in 2:12){
# if(some conditions) or calcs
trans.prob <- c( trans.prob , rbeta(1,12,1,trans.prob[i-1]) )
}
plot(trans.prob,type='l')

and don't really know how to update the prob distribution in
the
> rows.
in a loop:
trans.prob.matrix[i, ] <- some.update.fun
> If anyone has any thoughts, i would be grateful for sharing.
If you'd read the posting guide and contemplated some existential
questions like "is it OK if the question (even if it's not) sounds
like homework ?" and "what's this obsession with providing
minimal,
reproducible code ? " than maybe the smaRter than me on this list will
be more inclined to share their thoughts ...

Cheers
Elai

> Stivi
>
> --
> View this message in context:
http://r.789695.n4.nabble.com/Updating-a-Markov-Chain-tp4504156p4504156.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

Thank you for your insights,  to give a little more background on what i am
looking for:
-I am trying to handle large sets of data (between 20 000 to 50 000)
entries, but the number of distinct values is very low (maybe 20 - 30, 50 in
extreme cases).  Hence handling of data should be as automated as possible.
-The main objective of this is to obtain the afore mentioned transition
matrix in a usable form.
-The matrix will be updated later also.
Regards,
Stivi

--
View this message in context:
http://r.789695.n4.nabble.com/Updating-a-Markov-Chain-tp4504156p4507049.html
Sent from the R help mailing list archive at Nabble.com.