Hi all,
How can a multivariate Poisson time series be modeled? Aspects of glm,
forecast, dse and dynlm seem relevant but not quite complete--but hopefully
what I am missing is how to assemble them effectively. What I am looking to
do is model my dependent variable y_t as a Poisson family function of lags
of several independent variables and lags of y_t. I would like to include
all lags up through t-n where n is specified and greater than 1, and have
the lagged values weighted by a decay factor that increases exponentially
with each time step further back in time from t. The time delay n of course
I can construct by redefining my variables, but I'm still struggling with
what functions will do the other aspects... but I have to think that someone
has done this before.
glm does Poisson but not time series explicitly, thus nor exponential decay
Arima() from forecast does time series obviously but not multivariate
prediction or decaying lags
dse() does multivariate time series regression but not Poisson or decaying
lags (or does it?)
dynlm() does multivariate time series regression but not Poisson or decaying
lags
Suggestions?
--
View this message in context:
http://r.789695.n4.nabble.com/Multivariate-time-series-Poisson-with-delayed-lags-tp3068284p3068284.html
Sent from the R help mailing list archive at Nabble.com.
