feat: add support for more neural models#338
Merged
Conversation
Contributor
There was a problem hiding this comment.
Pull request overview
This PR expands timecopilot.models by adding additional auto-tuned NeuralForecast and MLForecast model wrappers, and by standardizing/supporting quantiles-based probabilistic outputs (while explicitly rejecting level) for these “Auto*” model families.
Changes:
- Added new NeuralForecast auto model wrappers:
AutoNBEATS,AutoDeepAR,AutoPatchTST, and enabledquantilessupport for the existing auto neural models viaMQLoss+QuantileConverter. - Added multiple new MLForecast auto model wrappers (e.g.,
AutoXGBoost,AutoCatboost, linear models, forests) and addedcatboost/xgboostas core dependencies. - Updated exports (
timecopilot.models.__init__) and test coverage for the expanded model set; added/updated changelog entries.
Reviewed changes
Copilot reviewed 8 out of 9 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
timecopilot/models/neural.py |
Adds new auto neural wrappers and routes probabilistic output through QuantileConverter + MQLoss. |
timecopilot/models/ml.py |
Adds new AutoMLForecast-based wrappers and a shared run_automlforecast_model helper for quantile outputs. |
timecopilot/models/__init__.py |
Re-exports new ML and neural models in the public package surface. |
tests/models/conftest.py |
Registers the new models in the test matrix with minimal configs for speed. |
tests/models/test_models.py |
Updates tests to reflect quantile support and the expanded auto-model set. |
pyproject.toml |
Adds catboost and xgboost to runtime dependencies. |
uv.lock |
Locks new dependencies (notably catboost, xgboost, graphviz updates). |
docs/changelogs/v0.0.25.md |
Adds new feature notes (but modifies an existing release file). |
docs/changelogs/v0.0.26.md |
Introduces a new changelog entry for quantile support improvements. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+4
to
+33
| from mlforecast.auto import ( | ||
| AutoCatboost as _AutoCatboost, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoElasticNet as _AutoElasticNet, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoLasso as _AutoLasso, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoLightGBM as _AutoLightGBM, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoLinearRegression as _AutoLinearRegression, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoMLForecast, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoRandomForest as _AutoRandomForest, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoRidge as _AutoRidge, | ||
| ) | ||
| from mlforecast.auto import ( | ||
| AutoXGBoost as _AutoXGBoost, | ||
| ) | ||
| from mlforecast.utils import PredictionIntervals | ||
|
|
||
| from .utils.forecaster import Forecaster, get_seasonality | ||
| from .utils.forecaster import Forecaster, QuantileConverter, get_seasonality |
Comment on lines
+1
to
+19
| ### Features | ||
|
|
||
| * **New neural models**: Added 3 new auto neural models powered by `neuralforecast`: `AutoNBEATS`, `AutoDeepAR`, and `AutoPatchTST`. All support `quantiles` for probabilistic forecasts trained with `MQLoss` and follow the same interface as the existing `AutoNHITS` and `AutoTFT`. | ||
|
|
||
| ```python | ||
| import pandas as pd | ||
| from timecopilot.models.neural import AutoDeepAR, AutoNBEATS, AutoPatchTST | ||
|
|
||
| df = pd.read_csv( | ||
| "https://timecopilot.s3.amazonaws.com/public/data/air_passengers.csv", | ||
| parse_dates=["ds"], | ||
| ) | ||
|
|
||
| model = AutoNBEATS() | ||
| fcst_df = model.forecast(df, h=12, quantiles=[0.1, 0.5, 0.9]) | ||
| ``` | ||
|
|
||
| * **New ML models**: Added 7 new auto ML models powered by `mlforecast`'s hyperparameter optimization: `AutoLinearRegression`, `AutoXGBoost`, `AutoRidge`, `AutoLasso`, `AutoElasticNet`, `AutoRandomForest`, and `AutoCatboost`. All models support `quantiles` for probabilistic forecasts via conformal prediction and follow the same interface as the existing `AutoLGBM`. | ||
|
|
Comment on lines
+341
to
+359
| if level is not None and quantiles is not None: | ||
| raise ValueError( | ||
| "You must not provide both `level` and `quantiles` simultaneously." | ||
| ) | ||
| if level is not None: | ||
| raise ValueError( | ||
| "Level is not supported for AutoNBEATS. " | ||
| "Please use `quantiles` instead." | ||
| ) | ||
|
|
||
| inferred_freq = self._maybe_infer_freq(df, freq) | ||
| qc = QuantileConverter(level=None, quantiles=quantiles) | ||
| loss = MQLoss(level=qc.level) if qc.level is not None else MAE() | ||
| if self.config is None: | ||
| config = _AutoNBEATS.get_default_config(h=h, backend="ray") | ||
| config["scaler_type"] = tune.choice(["robust"]) | ||
| else: | ||
| config = self.config | ||
| if self.backend == "optuna": |
Comment on lines
254
to
282
| def test_using_level(model): | ||
| level = [0, 20, 40, 60, 80] # corresponds to qs [0.1, 0.2, ..., 0.9] | ||
| df = generate_series(n_series=2, freq="D") | ||
| if model.alias in [ | ||
| _level_unsupported = { | ||
| "AutoLGBM", | ||
| "AutoLinearRegression", | ||
| "AutoXGBoost", | ||
| "AutoRidge", | ||
| "AutoLasso", | ||
| "AutoElasticNet", | ||
| "AutoRandomForest", | ||
| "AutoCatboost", | ||
| "AutoNHITS", | ||
| "AutoTFT", | ||
| "PatchTST-FM", | ||
| ]: | ||
| # These models do not support levels yet | ||
| "AutoNBEATS", | ||
| "AutoDeepAR", | ||
| "AutoPatchTST", | ||
| } | ||
| if model.alias in _level_unsupported: | ||
| # these models only support quantiles, not level | ||
| with pytest.raises(ValueError) as excinfo: | ||
| model.forecast( | ||
| df=df, | ||
| h=2, | ||
| freq="D", | ||
| level=level, | ||
| ) | ||
| assert "not supported" in str(excinfo.value) | ||
| assert "quantiles" in str(excinfo.value) | ||
| return |
Member
Author
Co-authored-by: AzulGarza <10517170+AzulGarza@users.noreply.github.com>
|
|
Contributor
Resolved the merge conflicts in a7fa6ba. The merge brings in origin/main (commits bebe988 and 63059d9) while keeping all changes from the current PR. Conflict resolutions:
|
Contributor
There was a problem hiding this comment.
Pull request overview
Adds additional NeuralForecast “auto” neural forecasters to the timecopilot.models surface area, along with tests and changelog updates.
Changes:
- Added new neural model wrappers:
AutoNBEATS,AutoDeepAR,AutoPatchTST. - Exported new models from
timecopilot.modelsand included them in the test matrix/fixtures. - Updated changelogs to document the new neural models.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
timecopilot/models/neural.py |
Adds NeuralForecast auto-model wrappers for NBEATS, DeepAR, PatchTST using the existing wrapper pattern. |
timecopilot/models/__init__.py |
Exposes the new neural models at the package level via imports and __all__. |
tests/models/test_models.py |
Updates tests to account for new neural auto aliases (min-length handling, quantile monotonicity exceptions, level unsupported list). |
tests/models/conftest.py |
Adds the new models to the parametrized test model fixture with lightweight configs. |
docs/changelogs/v0.0.26.md |
Documents the new neural models and provides a usage example. |
docs/changelogs/v0.0.25.md |
Removes an extra blank line in the features section. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+7
to
17
| ) | ||
| from neuralforecast.auto import ( | ||
| AutoNBEATS as _AutoNBEATS, | ||
| ) | ||
| from neuralforecast.auto import ( | ||
| AutoNHITS as _AutoNHITS, | ||
| ) | ||
| from neuralforecast.auto import ( | ||
| AutoPatchTST as _AutoPatchTST, | ||
| ) | ||
| from neuralforecast.auto import ( |
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
this pr adds support for more neural models