Open JustM57 opened 1 year ago
Yeah this has been brought up from time to time, but there's no active work on new modelling techniques at the moment. Some strategies are discussed here to modify the existing prophet model: https://github.com/facebook/prophet/issues/49
neuralprophet
might have an approach already, although I don't think it uses shared coefficients.
It's clear that we can build several models with independent trend, seasonality and coefficients for covariates. Is it possible to perform forecast for multilpe series with a single model? I mean have different trend seasonality, but shared(same) coefficients for covariates?