Kevin Chen

TODO: notebook example and update website for ts forecasting w panel datasets

Late to convo but I agree with @ZviBaratz here that this doesn't improves any existing functionality. The imported utils are not used anywhere throughout the project so why is there...

got it. will do

Currently, FLAML only supports single time-dependent variable for time series forecasting, as the learners only support univariate time series and do not take in exogenous regressors.

At the moment, we will be improving time series forecasting to support continuous features. In regards to categorical features, we ask that users treat it as hierarchical time series forecasting,...

The current PR will only support multivariate time series predictions for regression problems (case 1), but the columns can contain different continuous and/or discrete/category observation. In case two, the target...

If you are referring to sklearn regressor models, we do lag the label before training. See https://github.com/microsoft/FLAML/blob/9901156deec8eb2a603e2fbf74ca01ceac5e9b5c/flaml/model.py#L2050. We use hcrystalball's `_transform_data_to_tsmodel_input_format` function to do this.

oh I see the issue. should be a minor fix to `gpus=kwargs.get("gpu_per_trial", [0])`. Can you submit a quick pr that changes that?

yes thank you for the comment, we are experiencing issues with the new pytorch-lightning version (as discussed in the discord group)