Nils Lehmann

@adamjstewart I tried adressing your suggestions to finish this up. But wanted to get your thoughts on the following: There is actually a complication that I am not entirely sure...

In case 1 the normalization would only be available throughtthe Datamodule and you would have to implement it in the on after batch transform and it would not be available...

But in case 2 I actually need to compute it on the fly because the regression target range changes based on the range restriction.

It's not different from case 1. Since the target range can change and there are no defined train/test splits on the dataset class level, the target normalization needs to happen...

These cyclone dataset usually have a very imbalanced target distribution (there are few images of high hurricane categories) and the min feature values would allow the user to basically only...

Thank you for your feedback. Your proposed integration/change within the `RasterDataset` is a better solution, so I am working on that at the moment: ### Dataset - changing `_merge_files` I...

This last commit contains a proposal for changing `RasterDataset` according to your two proposed changes. And I added some pytests for the dataset but not yet for the samplers.

Given the complexity of supporting all the different time-series scenario tasks, my idea is to first begin by focusing on a single one that should already be quiet powerful for...

The above commits contain a new design proposal that samples consecutive sequences for time-series forecasting tasks. It contains the following new pieces: - `ForecastDataset`: inherits from `IntersectionDataset` with the idea...

Maybe it is useful to determine some terminology that can be used for models, datasets, samplers and tasks for the context of time-series prediction. These are just a few terms...