Any extension planned to simultaneous multiple time series forecasting?

[catubc]

Hello

This is an amazing package, thanks so much.

We have datasets that have > 1 time series that we'd like to building models from. For example, instead of 1D input [t0, t1, ...tn], we have [[t0,t1,...tn], [k0, k1, ... kn]]

Is there any extension planned for this? For clarity, the time series interact in very complex ways that are hard to disentangle - and we're hoping NNs can extract some of that structure for us.

We're also happy to contribute to making this happen - just need a bit of help/direction.

Thanks so much, catubc

[kashif]

@catubc I believe as in the time series blog posts, the datasets are a collection of time-series and a single shared model is trained over them...