Question about exogenous variables

yuqinie98 / PatchTST

An offical implementation of PatchTST: "A Time Series is Worth 64 Words: Long-term Forecasting with Transformers." (ICLR 2023) https://arxiv.org/abs/2211.14730

Apache License 2.0

1.37k stars 248 forks source link

Question about exogenous variables #88

Open RuohanRYAN opened 7 months ago

RuohanRYAN commented 7 months ago

I noticed that only batch_x were passed in the model during training. Are the exogenous variables related to Datetimes not modeled in this architecture or it there a way to include them?

pinyin commented 4 months ago

The neuralforecast's PatchTST implemention also declares itself as not supporting exogenous variables FYI. I'm not a professional in ML, just being courious that whether it's an incomplete implemention or a limition by design.