NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI
PyPI, #3253 most downloaded on PyPI
Time series forecasting suite using statistical models
Latest stable
2.1.1
Last release 4 months ago
16 Jul 2026
Release timing varies
gaps range from 2 weeks to 9 months
Nearly every release is documented
notes for 40 of 43 stable releases
Nothing withdrawn
no release was ever pulled
5 years old
43 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
[CHORE] Publish Python 3.14 wheels @nasaul
[FEAT] Deprecate Numba @elephaint
pyproject.toml and other workflows to use uv @deven367 (#1080)blog @deven367 (#1076)Your coding agent can read these notes before it upgrades. Set up the MCP server →
Documentation improvement @nasaul
df arg from constructor @deven367 (#1030)FutureWarning: 'H' is deprecated and will be removed in a future version, please use 'h' instead. @vaidatascientist
fix(nelder-mead): argsort before early exit @jmoralez
mean with fitted in doc string of `predict_in_sampl… @Gunnvant (#978)breaking: remove deprecated behavior @jmoralez
fix: chunk series in parallel forecast @jmoralez
fix: bump build to include submodules in sdist @jmoralez
fix: reduce overhead in parallel forecast @jmoralez
fix matrix product for arima var_coef @jmoralez
Fix allowdrift and allowmean in non-stepwise AutoARIMA @manuel-calzolari
fix: parallel custom cols @AzulGarza
fix fitted values for sparse models @jmoralez
AutoTBATS, experiment, and minor issues @MMenchero
use future instead of deprecation warnings @jmoralez
improve deprecation error messages @jmoralez
requirements.txt @akmalsoliev (#666)take shallow copy on dataframe processing and fix get_cmap deprecation @jmoralez
Republish of the 1.6.0 release from August 23rd 2023, since it disappeared from github.
settings.ini @akmalsoliev (#499)unique_id @nickto (#473)ets_f lower and upper arguments @kschmaus (#437)nbdev @akmalsoliev (#449)[FEAT] ARIMA model (no auto version) in https://github.com/Nixtla/statsforecast/pull/383
Now you can pre-train a model and use new data to make forecasts through the forward method. Supported models:
languages in https://github.com/Nixtla/statsforecast/pull/356Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.4.0...v1.5.0
feat: Added prediction intervals for insample and ETS models in https://github.com/Nixtla/statsforecast/pull/328
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.3.2...v1.4.0
[FIX] Improvements to StatsForecast's plot method in https://github.com/Nixtla/statsforecast/pull/312
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.3.1...v1.3.2
[FEAT] Add plot method to StatsForecast class in https://github.com/Nixtla/statsforecast/pull/305
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.3.0...v1.3.1
[FIX] Use conda env for ray tests in https://github.com/Nixtla/statsforecast/pull/297
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.2.1...v1.3.0
[FEAT]: Add fallback model to cross validation in https://github.com/Nixtla/statsforecast/pull/289
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.2.0...v1.2.1
[FEAT] MSTL model n https://github.com/Nixtla/statsforecast/pull/284
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.1.3...v1.2.0
[FEAT] Add progress bar for sequential tasks in https://github.com/Nixtla/statsforecast/pull/280
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.1.2...v1.1.3
[FEAT] Improve navbar docs in https://github.com/Nixtla/statsforecast/pull/262
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.1.1...v1.1.2
[FEAT] Add Distributed post in https://github.com/Nixtla/statsforecast/pull/257
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.1.0...v1.1.1
[FIX] License in https://github.com/Nixtla/statsforecast/pull/191
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v1.0.0...v1.1.0
[BREAKING CHANGE] SKLearn syntax in https://github.com/Nixtla/statsforecast/pull/184 * Full Changelog: https://github.com/Nixtla/statsforecast/compare…
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.7.1...v1.0.0
[FEAT] Fitted df returns in-sample values in https://github.com/Nixtla/statsforecast/pull/158
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.7.0...v0.7.1
[BREAKING CHANGE] Fitted Values Computation in https://github.com/Nixtla/statsforecast/pull/137
mean and fitted values.Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.6.0...v0.7.0
[BREAKING CHANGE] Deprecate python3.6 in https://github.com/Nixtla/statsforecast/pull/146
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.5.6...v0.6.0
[DOCS] Typo fixes by @ryanrussell in https://github.com/Nixtla/statsforecast/pull/117
n_windows argument for cross_validation method by @FedericoGarza in https://github.com/Nixtla/statsforecast/pull/131Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.5.5...v0.5.6
ARIMA level/quantile compatibility, missing nbdev_flow, protected gif by @kdgutier in https://github.com/Nixtla/statsforecast/pull/102
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.5.4...v0.5.5
feat: add issues template by @FedericoGarza in https://github.com/Nixtla/statsforecast/pull/93
Full Changelog: https://github.com/Nixtla/statsforecast/compare/v0.5.3...v0.5.4
summary method for the AutoARIMA class requested in #31.
summary method for the AutoARIMA class requested in #31.AutoARIMA fitted model, requested in #83.croston_sba #88 fixed in #89.Added predict_in_sample method for AutoARIMA.
predict_in_sample method for AutoARIMA.Now: Good Ol' sklearn syntax with model = AutoARIMA(); model.fit(y); model.predict(10).
model = AutoARIMA(); model.fit(y); model.predict(10).Inclusion of prediction intervals for auto_arima.
prediction intervals for auto_arima.statsforecast is now installable from conda-forge (conda install -c conda-forecast statsforecast, thanks to @sugatoray).Inclusion of exogenous variables for auto_arima.
exogenous variables for auto_arima.StatsForecast class now handles exogenous variables.exogenous variables.Nothing published for this version
Nothing published for this version
Nothing published for this version
One column per quarter.