Merge pull request #518 from jdb78/feature/better-time-series-docs

feature/better-time-series-docs

Author: jdb78

.github/workflows/automerge.yml

@@ -3,6 +3,10 @@ on:
   pull_request_target:
     branches: [master]
 
+  check_suite:
+    types:
+      - completed
+
 jobs:
   autoapprove:
     name: Auto Approve a PR by dependabot
@@ -24,3 +28,5 @@ jobs:
           GITHUB_TOKEN: "${{ secrets.GITHUB_TOKEN }}"
           MERGE_LABELS: dependencies
           MERGE_METHOD: rebase
+          MERGE_RETRY_SLEEP: 600000
+          MERGE_RETRIES: 100

CHANGELOG.md

@@ -1,12 +1,35 @@
 # Release Notes
 
-## UNRELEASED
+## v0.9.0 Simplified API (04/06/2021)
+
+### Breaking changes
+
+- Removed `dropout_categoricals` parameter from `TimeSeriesDataSet`.
+  Use `categorical_encoders=dict(<variable_name>=NaNLabelEncoder(add_nan=True)`) instead (#518)
+- Rename parameter `allow_missings` for `TimeSeriesDataSet` to `allow_missing_timesteps` (#518)
+- Transparent handling of transformations. Forward methods should now call two new methods (#518):
+
+  - `transform_output` to explicitly rescale the network outputs into the de-normalized space
+  - `to_network_output` to create a dict-like named tuple. This allows tracing the modules with PyTorch's JIT. Only `prediction` is still required which is the main network output.
+
+  Example:
+
+  ```python
+  def forward(self, x):
+      normalized_prediction = self.module(x)
+      prediction = self.transform_output(prediction=normalized_prediction, target_scale=x["target_scale"])
+      return self.to_network_output(prediction=prediction)
+  ```
 
 ### Fixed
 
 - Fix quantile prediction for tensors on GPUs for distribution losses (#491)
 - Fix hyperparameter update for RecurrentNetwork.from_dataset method (#497)
 
+### Added
+
+- Improved validation of input parameters of TimeSeriesDataSet (#518)
+
 ## v0.8.5 Generic distribution loss(es) (27/04/2021)
 
 ### Added

docs/source/faq.rst

@@ -32,7 +32,7 @@ Creating datasets
 
 * **It takes very long to create the dataset. Why is that?**
 
-  If you set ``allow_missings=True`` in your dataset, the creation of an index
+  If you set ``allow_missing_timesteps=True`` in your dataset, the creation of an index
   might take far more time as all missing values in the timeseries have to be identified.
   The algorithm might be possible to speed up but currently, it might be faster for you to
   not allow missing values and fill them yourself.