I've found a fair bit of research on using Metropolis-Hastings on time series data. In everything I've read so far the focus is on identifying the correlation. I don't really need to do that. I already know the correlation boundaries. What I need to be able to do is build those into my model so I can adjust for them.
In the mean time, I think just by pulling multiple blocks I can smooth out most queries pretty well.
No comments:
Post a Comment