Statsforecast

Republish of the 1.6.0 release from August 23rd 2023, since it disappeared from github.

New Features

- Conformal Prediction kvnkho (592)
- Adding levels to distributed backends kvnkho (581)
- [FEAT] Add ConstantModel and ZeroModel FedericoGarza (568)
- [FEAT] Add NaNModel FedericoGarza (567)
- [FEAT] Add conformal intervals Theta family FedericoGarza (501)
- [FEAT] Add conformal intervals for CES FedericoGarza (500)
- [FEATURE] Polars support akmalsoliev (448)
- [FEAT] Add conformal intervals to arima family FedericoGarza (488)
- [FEAT] Add conformal intervals to StatsForecast class FedericoGarza (487)
- [FEAT] Add conformal prediction for AutoARIMA FedericoGarza (486)
- [FEAT] Return plot object FedericoGarza (465)
- [FEAT] Add stl_kwargs to MSTL FedericoGarza (462)
- [FEAT] Support pandas 2.0 changes FedericoGarza (456)

Breaking Change

- Reducing StatsForecast Size kvnkho (600)

Bug Fixes

- take shallow copy on dataframe processing and fix get_cmap deprecation jmoralez (617)
- fix arima max order params jmoralez (613)
- Fix iteration range in non-stepwise AutoARIMA manuel-calzolari (601)
- [Core] Fixed RuntimeWarning Generated by _get_cols (538) taniishkaaa (563)
- [FIX] Unnecessary datetime column conversion akmalsoliev (558)
- [FIX] Accommodated switch to jupyter-lab akmalsoliev (511)
- [FIX] Polars hotfix akmalsoliev (503)
- [FIX] Added polars to `settings.ini` akmalsoliev (499)
- [FIX] HoltWinters forecasts (weekly seasonality) FedericoGarza (483)
- [FIX] Consider correct seasonality for exp smoothing FedericoGarza (474)
- Remove unused levels from categorical `unique_id` nickto (473)
- [FIX] Add protection ETS zero division error FedericoGarza (470)
- [FIX] allow period=1 using mstl FedericoGarza (463)
- [FIX] ets forbidden component combinations FedericoGarza (461)
- [FIX] Different results between forecast and fit/predict MSTL FedericoGarza (446)
- Stop using mutable defaults for `ets_f` `lower` and `upper` arguments kschmaus (437)
- [FIX] Distributed behaviour with exogenous variables FedericoGarza (427)

Documentation

- Adding details to Conformal Prediction docs kvnkho (607)
- Tutorial for Conformal Prediction kvnkho (597)
- [FIX] SeasonalNaive docs nelsoncardenas (588)
- Fix incorrect parameter name in How-To-Guides yibenhuang (584)
- Changing Load Forecasting Data Souce kvnkho (572)
- [DOCS] Adding GARCH and ARCH to index kvnkho (571)
- Fix broken doc links andrewgross (566)
- Fixing broken links kvnkho (559)
- Updated the document to reflect the deprecation of ETS in favor of AutoETS (319) taniishkaaa (561)
- [DOC] renamed files for order mergenthaler (554)
- Update nbs/ FedericoGarza (548)
- [FIX] Restructure how-to guides FedericoGarza (547)
- [DOCS] AutomaticForecasting mergenthaler (545)
- Updating Distributed Documentation kvnkho (541)
- Update nbs/ FedericoGarza (546)
- [FEAT] New docs structure FedericoGarza (534)
- [DOCS] Polars documentation akmalsoliev (527)
- Update nbs/docs/contribute/ file FedericoGarza (544)
- Update nbs/docs/contribute/ file FedericoGarza (543)
- Update nbs/ file FedericoGarza (542)
- Update CONTRIBUTING.md file FedericoGarza (533)
- [FEAT] Add mlforecast to ensemble example FedericoGarza (502)
- [FIX] Link end to end pipeline FedericoGarza (477)
- Update README.md mergenthaler (468)
- [DOCS] Added more instructions on `nbdev` akmalsoliev (449)
- [DOCS] Hide utils fns from core FedericoGarza (429)
- Fix naive model description shagn (268)

Enhancement

- check for level when prediction_intervals are set jmoralez (615)
- raise informative error when series are too short for cross_validation jmoralez (610)
- Add release drafter FedericoGarza (514)
- Add issue template files FedericoGarza (513)
- Add issue template FedericoGarza (512)

What's Changed

Features

New models
* [FEAT] ARIMA model (no auto version) in https://github.com/Nixtla/statsforecast/pull/383
* [FEAT] AutoRegressive model in https://github.com/Nixtla/statsforecast/pull/387
* [FEAT] GARCH and ARCH models in https://github.com/Nixtla/statsforecast/pull/403

New functionality

Forward methods

Now you can pre-train a model and use new data to make forecasts through the `forward` method. Supported models:
* [FEAT] Add forward method to Theta models in https://github.com/Nixtla/statsforecast/pull/362
* [FEAT] Add forward method to ETS models in https://github.com/Nixtla/statsforecast/pull/363
* [FEAT] Add forward method to AutoCES class in https://github.com/Nixtla/statsforecast/pull/364
* [FEAT] Add forward method to MSTL model in https://github.com/Nixtla/statsforecast/pull/369
* [FEAT] Add forward method to AutoARIMA (ARIMA and AutoRegressive) in https://github.com/Nixtla/statsforecast/pull/368

Misc

* [FEAT] Add alias argument to models (fit the same instance of models with different names) in https://github.com/Nixtla/statsforecast/pull/357
* [FEAT] Add cross-validation without refit (using the forward method) in https://github.com/Nixtla/statsforecast/pull/370
* [FEAT] Allow seasonality greater than 24 for ETS in https://github.com/Nixtla/statsforecast/pull/384
* [FEAT] Allow passing fixed coefficients for Arima in https://github.com/Nixtla/statsforecast/pull/386
* [FEAT] AutoCES prediction intervals in https://github.com/Nixtla/statsforecast/pull/394 (now StatsForecast is fully probabilistic)
* [FEAT] Add cla workflow in https://github.com/Nixtla/statsforecast/pull/351
* [FEAT] Add pypi downloads badge in https://github.com/Nixtla/statsforecast/pull/352
* [FEAT] Ignore jupyter notebooks as part of `languages` in https://github.com/Nixtla/statsforecast/pull/356
* [FEAT] Add nbdev merge to git attributes in https://github.com/Nixtla/statsforecast/pull/365
* [FEAT] Add citation in https://github.com/Nixtla/statsforecast/pull/366
* [FEAT] Update table of models in https://github.com/Nixtla/statsforecast/pull/396

Experiments

* [FEAT] Add M5 and M4-Daily experiments (Amazon Forecast) in https://github.com/Nixtla/statsforecast/pull/332
* [FEAT] Test recover M3 performance in https://github.com/Nixtla/statsforecast/pull/385
* [FEAT] BigQuery comparison in https://github.com/Nixtla/statsforecast/pull/421
* [FEAT] Experiments for ETS prediction intervals for multiple confidence levels in https://github.com/Nixtla/statsforecast/pull/377
* [FEAT] Add M3 experiment in https://github.com/Nixtla/statsforecast/pull/348
* [FEAT] Add a test ensuring the m3 performance is recovered in less than two minutes in https://github.com/Nixtla/statsforecast/pull/388

Tutorials

* [FEAT] Improved intermittent data nb in https://github.com/Nixtla/statsforecast/pull/359
* [FEAT] Add statistical and neural methods tutorial in https://github.com/Nixtla/statsforecast/pull/399
* [FEAT] Improve anomaly detection nb in https://github.com/Nixtla/statsforecast/pull/338
* [FEAT] GARCH and ARCH models tutorial in https://github.com/Nixtla/statsforecast/pull/418
* [FEAT] Improved notebook on prediction intervals in https://github.com/Nixtla/statsforecast/pull/358
* [FEAT] Improved notebook on exogenous regressors in https://github.com/Nixtla/statsforecast/pull/392
* [FEAT] Improve documentation in https://github.com/Nixtla/statsforecast/pull/376

Fixes
* [FIX] Exponential Smoothing description in https://github.com/Nixtla/statsforecast/pull/346
* [FIX] Changed dataset and model to make example easier to follow in https://github.com/Nixtla/statsforecast/pull/345
* [FIX] Readme M3 typo in https://github.com/Nixtla/statsforecast/pull/350
* [FIX] Delete CLA.yml in https://github.com/Nixtla/statsforecast/pull/355
* [FIX] Broken link in https://github.com/Nixtla/statsforecast/pull/360
* [FIX] Clean aws nbs in https://github.com/Nixtla/statsforecast/pull/361
* [FIX] Add correct link to hierarchicalforecast by https://github.com/Nixtla/statsforecast/pull/372
* [FIX] Recover table-based documentation (core nb, compatible with docstrings) in https://github.com/Nixtla/statsforecast/pull/374
* [FIX] update sklearn -> scikit-learn in https://github.com/Nixtla/statsforecast/pull/375
* [FIX] Ray CI in https://github.com/Nixtla/statsforecast/pull/381
* [FIX] Links and typos in documentation in https://github.com/Nixtla/statsforecast/pull/390
* [FIX] Correct evaluation using Winkler score by MMenchero in https://github.com/Nixtla/statsforecast/pull/395
* [FIX] Recover plots prediction intervals tutorial in https://github.com/Nixtla/statsforecast/pull/398
* [FIX] Use https links instead of s3 uris (stat-neural tutorial) in https://github.com/Nixtla/statsforecast/pull/400
* [FIX] New nbdev clean behaviour in https://github.com/Nixtla/statsforecast/pull/412
* [FIX] Model imports in https://github.com/Nixtla/statsforecast/pull/408

New dependencies

* [FEAT] plotly-resampler as plotting engine in https://github.com/Nixtla/statsforecast/pull/354
* [FEAT] Move Fugue to core dependency in https://github.com/Nixtla/statsforecast/pull/419

New Contributors
* jvdd made their first contribution in https://github.com/Nixtla/statsforecast/pull/354
* Roymprog made their first contribution in https://github.com/Nixtla/statsforecast/pull/390
* nelsoncardenas made their first contribution in https://github.com/Nixtla/statsforecast/pull/408

**Full Changelog**: https://github.com/Nixtla/statsforecast/compare/v1.4.0...v1.5.0

What's Changed
* feat: Added prediction intervals for insample and ETS models in https://github.com/Nixtla/statsforecast/pull/328
* [FEAT] Add plot anomalies option in https://github.com/Nixtla/statsforecast/pull/341
* [DOCS] Improve README and docs page index in https://github.com/Nixtla/statsforecast/pull/344


**Full Changelog**: https://github.com/Nixtla/statsforecast/compare/v1.3.2...v1.4.0

What's Changed
* [FIX] Improvements to StatsForecast's plot method in https://github.com/Nixtla/statsforecast/pull/312
* [FEAT] Add plotly as engine to StatsForecast's plot method in https://github.com/Nixtla/statsforecast/pull/313
* [FEAT] Add autowidth to plotly engine in https://github.com/Nixtla/statsforecast/pull/314
* feat: add new documentation in https://github.com/Nixtla/statsforecast/pull/317
* [FIX] ETS for inttermitent series in https://github.com/Nixtla/statsforecast/pull/315
* [FIX] Theta for intermittent series in https://github.com/Nixtla/statsforecast/pull/316
* [FEAT] Rename ETS to AutoETS in https://github.com/Nixtla/statsforecast/pull/318
* [FEAT] Change library to newest black formatting in https://github.com/Nixtla/statsforecast/pull/320
* [FIX] Add new plot method to mstl example in https://github.com/Nixtla/statsforecast/pull/324
* [FIX] Build docs for Theta model in https://github.com/Nixtla/statsforecast/pull/322
* [FIX] Isolate elements for all subplots plotly in https://github.com/Nixtla/statsforecast/pull/323
* Fix/multiple seas docs in https://github.com/Nixtla/statsforecast/pull/325
* [FEAT] Add mstl experiment in https://github.com/Nixtla/statsforecast/pull/326
* [FIX] Prevent futurewarning series indexing in https://github.com/Nixtla/statsforecast/pull/327
* Fix sidebar in https://github.com/Nixtla/statsforecast/pull/331
* feat: Improved tutorial on Cross-Validation in https://github.com/Nixtla/statsforecast/pull/333
* Feat/improve prediction intervals in https://github.com/Nixtla/statsforecast/pull/336
* fix: Improved AutoARIMA plot in https://github.com/Nixtla/statsforecast/pull/334
* docs: ERCOT electricity demand peak forecasting in https://github.com/Nixtla/statsforecast/pull/335
* docs: fix peak demand plot in https://github.com/Nixtla/statsforecast/pull/339

New Contributors
* cchallu made their first contribution in https://github.com/Nixtla/statsforecast/pull/335

**Full Changelog**: https://github.com/Nixtla/statsforecast/compare/v1.3.1...v1.3.2

What's Changed
* [FEAT] Add plot method to StatsForecast class in https://github.com/Nixtla/statsforecast/pull/305
* [FEAT] New Issues Templates in https://github.com/Nixtla/statsforecast/pull/307
* [FIX] make logging config local to package in https://github.com/Nixtla/statsforecast/pull/275
* [FIX] Error when ds column is object in https://github.com/Nixtla/statsforecast/pull/309

New Contributors
* JeroenPeterBos made their first contribution in https://github.com/Nixtla/statsforecast/pull/275

**Full Changelog**: https://github.com/Nixtla/statsforecast/compare/v1.3.0...v1.3.1

What's Changed
* [FIX] Use conda env for ray tests in https://github.com/Nixtla/statsforecast/pull/297
* [FIX] Source code broken links in https://github.com/Nixtla/statsforecast/pull/293
* [FIX] Sparse models with zero-valued time series in https://github.com/Nixtla/statsforecast/pull/294
* [FIX] Add explicit optional argument (PEP-484) in https://github.com/Nixtla/statsforecast/pull/301
* [FIX] SeasonalNaive in https://github.com/Nixtla/statsforecast/pull/302
* [FEAT] Add exogenous variables to fugue's backend in https://github.com/Nixtla/statsforecast/pull/300
* [FEAT] Add Theta methods in https://github.com/Nixtla/statsforecast/pull/299
* [FEAT] Add MSTL example and comparison in https://github.com/Nixtla/statsforecast/pull/295
* [FEAT] Add backend argument to StatsForecast class in https://github.com/Nixtla/statsforecast/pull/303


**Full Changelog**: https://github.com/Nixtla/statsforecast/compare/v1.2.1...v1.3.0