Multivariate time-series RNN using Tensorflow. Is this possible with an LSTM cell or similar?

Question

I am looking for examples of how to build a multivariate time-series RNN using Tensorflow. Is this possible with an LSTM cell or similar?

e.g. the data might look something like this:

Time,A,B,C,...

0,3.5,4.5,7.7,...

1,2.1,6.4,8.2,...

...

Any help much appreciated. Thanks, John

Answer 1

It depends on exactly what you mean, but yes, it should be possible. If you write more specifically how exactly your input and target data looks like, somebody may be able to help. You can generally have sequential continuous or categorical input data and sequential continuous or categorical output data or a mix of those. I would suggest you look at the tutorials and try out a few things, then ask again here.