Hi,
Does R have any functions implementing such multinomial regression:
(S_t^A,S_t^B)~MN(N_t-Y_{t-1},P_t^A,P_t^B)
where MN(n,p_1,p_2) is multinomial distribution with parameters n, p_1, p_2.
Here P_t^A and P_t^B are nonlinear functions from predictor variables and
parameters which need to be estimated.
Here A and B are used for notation, they are not parameters.
My second question is about nls capabilities. Can I use lagged response
variable in right side of formula specified in nls. Like this:
nls(y~a*f(B(y)),data=data,start=list(a=1))
where B is lag operator. Or should I just use lagged response variable as
predictor variable:
nls(y~a*f(x),data=data1,start=list(a=1))
data1<-data.frame(y=y,x=c(0,y[1:(length(y)-1)]))
here f is arbitrary function.
Thank you in advance for any answers.
Sincerely,
Vaidotas Zemlys
PS I'm resending this mail, since it seems that it didn't get through
the
first time. I am sorry, if you got 2 messages.
