Multivariate time-series regression in PyG

[chara995]

Hello,

I am a beginner in PyG, so excuse me in advance if my question is trivial.

I have six sensors placed in a specific geometrical layout. Each sensor measures two variables (displacement, speed). For every sensor, I have a time series of measurements.

I want to create a surrogate model where each sensor is a node with two features to forecast displacement and speed in the future.

I saw the Pytorch geometric temporal library, which sounds ideal for this work.

Could I do the same in PyG by implementing some "high-dimensional" features of the nodes? So passing inside the node feature the appropriate sliding window I need for the time regression?

Thank you so much!