During the Autoregressive and Normal Emission Hidden Markov Models video, @RavinKumar mentioned the applicability of HMMs to time series data where a different forecasting model could be applied depending on the current state. Does anyone know where can I find more information about it? I’m interested in cases where a time series with trend and weekly/yearly seasonality has trend changes. Something like what Prophet does for Trend Changepoints, but taking advantage of what HMMs and Dynamax can offer.
Any hints or references with code are highly appreciated.