NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #3126 most downloaded on PyPI
Forecasting utilities
Last release 5 months ago
27 Apr 2026
Release timing varies
gaps range from 1 weeks to 7 months
Nearly every release is documented
notes for 55 of 56 stable releases
Nothing withdrawn
no release was ever pulled
3 years old
57 releases · first in 2023
One column per quarter.
[FEAT] Fast path for sorted backtest splits using boundary indexing @janrth
Enhanced Scaled Metric Computation and Loss Function Support @elephaint , @nasaul
What's new:
Bug fixes:
Documentation:
Nothing published for this version
Nothing published for this version
Add Scaled QL and Scaled MQL @marcozanotti
fill_gaps @AlexMilanor (#154)fix(pandas): use arrays for values and indices in time_features @elephaint
deps(pandas): address frequency alias deprecations in generate_series @jmoralez
Add MSSE and RMSSE losses @marcozanotti
enh: ignore pandas perf warnings on unvectorized offsets @jmoralez
fix: ensure numpy in fill_gaps bounds @jmoralez
add preprocessing.id_time_grid @jmoralez
preprocessing.id_time_grid @jmoralez (#129)feat: add forecast bias metric @jmoralez
feat: add future_exog_to_historic function @jmoralez
support single ax object in plot_series @jmoralez
support returning multiple time features from function @jmoralez
add time_features to feature engineering @jmoralez
add agg_fn to evaluate @jmoralez
support lists in assign_columns @jmoralez
support timezones for pandas in fill_gaps @jmoralez
fix timestamp check in time_ranges for timestamps with timezone @jmoralez
faster counts_by_id for pandas @jmoralez
improve sorting when ids are strings @jmoralez
support A as yearly frequency in fill_gaps by @jmoralez in https://github.com/Nixtla/utilsforecast/pull/81
support quarters in fill_gaps @jmoralez
fix counts_by_id sorting for pandas @jmoralez
Make regular expression into raw string @jlopezpena
fix scaled_crps for pandas @jmoralez
support hours in fill_gaps @jmoralez
address deprecation warnings @jmoralez
fix probabilistic losses @jmoralez
Full Changelog: https://github.com/Nixtla/utilsforecast/compare/v0.0.27...v0.1.0
add drop_columns to address polars DeprecationWarning @jmoralez
check for pyarrow data types @jmoralez
address polars Series.dtype deprecation @jmoralez
## Dependencies - polars updates @jmoralez
rename polars value_counts output @jmoralez
add add_insample_levels function @jmoralez
set orient in pl.from_numpy @jmoralez
add colnames as arguments to cv_times @jmoralez
add make_future_dataframe and anti_join functions by @jmoralez in https://github.com/Nixtla/utilsforecast/pull/45
add validate_freq function @jmoralez
handle month ends for polars in time_ranges @jmoralez
handle month ends for polars in offset_times @jmoralez
support polars dataframe in fill_gaps @jmoralez
add ensure_time_dtype and handle pandas nullable dtypes in validate_format @jmoralez
redirect to mintlify docs @jmoralez
add match_categories argument to vertical_concat @jmoralez
add more functions to processing @jmoralez
dont transform when scale is zero @jmoralez
sort ids when computing sizes @jmoralez
## New Features - extend processing @jmoralez
fix minmax scaler denominator @jmoralez
add target transformations @jmoralez
fix plotting column selection by @jmoralez in https://github.com/Nixtla/utilsforecast/pull/12
handle irregular frequencies in fill_gaps @jmoralez
add evaluation and vectorize losses @jmoralez
rename plot to plot_series @jmoralez
This is the first release, it includes the following modules:
This is the first release, it includes the following modules:
generate_series: used for generating synthetic time seriesplot: visualization of series' history and model predictions along with confidence intervals and anomalies.fill_gaps: helps with filling missing dates in the dataframe.Your coding agent can read these notes before it upgrades. Set up the MCP server →