[R] Markov chain resources and questions

Paul Meagher paul at datavore.com
Sun Oct 19 16:02:35 CEST 2003


Can someone give me a pointer to where I should be looking for markov chain
resources in R?

Longer term, I am also interested in the question of whether explanatory
variables can coupled to a probability transition matrix to assist in
predicting the next state that a an object/system will go into.  I'm
imagining that this gets kind of ugly when you have nominal data (i.e., the
next state) that you are trying to predict using a transition matrix and you
want to try to boost your predictive power by incorporating other regressor
variables.  Can this be done, how, and is there something in R that does
this?

Another issue is how to assess the potential usefulness of the probability
transition matrix and the corresponding frequency matrix.  If your frequency
matrix only has a few observations in each cell then it is not too useful
for predictive purposes.  Also, if the probabilties are close to .50 in all
the cells, again it is not useful because it is not "informative" about the
next state.  Is their research that speaks to the issue of assessing the
"utility" or "informativeness" of the transition matrix for predictive
purposes.

Regards,
Paul Meagher




More information about the R-help mailing list