I wonder if extracting a time series per estimated parameter is possible after fitting. I did my fitting for 100 days in real cases. Summary of fitting gives me mean, st… value for each parameter and convert the fitting to a data frame representing 2000 value for 4 chains. I am looking for a time series (100 days) for each parameter.
fit <-sampling(model, data, iter=1000,seed=4)
m ← as.matrix(fit)
? extract beta for 100 days
Thanks in advance
I’m not sure what you mean by a time series for each parameter. If there is a latent parameter for each point you should have 100 parameters. If not, there isn’t a time-series of parameters per se.
Can you share your model and say what it is you’re trying to fit? Or just what kind of time series you’re talking about? For example, an AR(1) model is going to have a likelihood statement like this:
y[2:T] ~ normal(a + b * y[1:T-1], sigma);
so there are only two parameters,