LucaButera

SpatioTemporalDataset.from_dataset does not accept the transform parameter.

1

Issue: The convenience method SpatioTemporalDataset.from_dataset does not accept any transform parameter, so it either has to be set manually after instantiation or the user must use the standard constructor. Proposed...

ScalerModule masks Scaler transform

1

**The issue:** The current ScalerModule implementation inherits the bias and scale parameter from the given Scaler and then uses [its own transform implementation](https://github.com/TorchSpatiotemporal/tsl/blob/d771cbd8a811fdad924c7d7ca7d5d364e8641958/tsl/data/preprocessing/scalers.py#L442C2-L460C65). This is not transparent to the user,...

Added a t_dim parameter to Masked Metric, used to specify the time dimension.

https://github.com/TorchSpatiotemporal/tsl/blob/9a5cee97123cb904763e922f3825ac236e0b6dc9/tsl/metrics/torch/pinball_loss.py#L30 In pinball_loss the PyTorch Lightning Metric deprecated parameter `compute_on_step` is used. See Lightning-AI/torchmetrics#789 It should be removed accordingly. In general, it may make sense to update compatibility to Lightning...

Add parameter to specify which is the time dimension in MaskedMetric

1

https://github.com/TorchSpatiotemporal/tsl/blob/154cccf6a07c5cc0558569f4faa18928f3755603/tsl/metrics/torch/metric_base.py#L112-L122 In the referenced snippet, MaskedMetric's update function assumes the time dimension is the second. However, this leads to adding an unnecessary dummy batch dimension when we represent a batch...

https://github.com/TorchSpatiotemporal/tsl/blob/39270b72858cc33edb775e1602e4e2a178d42c13/tsl/data/batch.py#L76C5-L76C25 Missing "keys" method call. Easily fixed by adding parentheses.