Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.
MIT License
Bot releases are hidden (Show)
scaling
to the Prophet()
instantiation. Allows minmax
scaling on y
instead ofabsmax
scaling (dividing by the maximum value). scaling='absmax'
by default, preserving theholidays_mode
to the Prophet()
instantiation. Allows holidays regressors to haveholidays_mode
takes the same value as seasonality_mode
Prophet
object: preprocess()
and calculate_initial_params()
. Thesey
scaling, creating fourier series, regressor scaling,extra_output_columns
to cross_validation()
. The user can specify additional columnspredict()
to include in the final output alongside ds
and yhat
, for example extra_output_columns=['trend']
. Credits to @dchiang00hdays
module was deprecated last version and is now removed.Full Changelog: https://github.com/facebook/prophet/compare/v1.1.4...1.1.5
Published by tcuongd over 1 year ago
holidays
package for country holidays. Credits to @arkid15r in https://github.com/facebook/prophet/pull/2379
holidays
package, and removes reliance on unmaintained manual holidays entries in hdays.py
. Importing from the prophet.hdays
module has been deprecated and the module will be removed in the next release.holidays
package so will not be added automatically with .add_country_holidays()
. These can be added manually instead, see examples here.Published by tcuongd over 1 year ago
Full Changelog: https://github.com/facebook/prophet/compare/v1.1.2...v1.1.3-patched
Published by tcuongd over 1 year ago
.predict()
by up to 10x by removing intermediate DataFrame creations. Credits to @orenmatar (https://github.com/facebook/prophet/pull/2299)train()
and predict()
pipelines. Credits to @yoziru (https://github.com/facebook/prophet/pull/2334)construct_holiday_dataframe()
holidays
data based on holidays version 0.18..tar.gz
to install from source, or .tgz
for the macOS binary.Published by tcuongd about 2 years ago
predict()
function via vectorization of future draws. Details here. Credits to @orenmatar for the original blog post and @winedarksea for the implementation.
predict()
now has a new argument, vectorized
, which is true by default. You should see speedups of 3-7x for predictions, especially if the model does not use full MCMC sampling. When using growth='logistic'
with mcmc_samples > 0
, predictions may be slower, and in these cases you can fall back to the original code by specifying vectorized=False
.cmdstanpy
minimum version is now 1.0.4prophet.__version__
now returns the correct version.Published by tcuongd over 2 years ago
pystan==2.19.1.1
, which is no longer maintained. cmdstanpy
is now the sole stan backend. Credits to @WardBrian @akosfurton @malmashhadani-88rolling_mean_by_h
function used to calculate cross validation performance metrics. Credits to @RaymondMcTholidays
package version 0.13.Published by bletham over 3 years ago
Published by bletham over 3 years ago
Published by seanjtaylor over 4 years ago
holidays
and pandas
packages.cmdstanpy
backend now available in PythonPublished by bletham over 5 years ago
Published by seanjtaylor almost 6 years ago
Published by seanjtaylor over 6 years ago
Published by bletham almost 7 years ago
Published by bletham about 7 years ago
Published by seanjtaylor over 7 years ago
Published by bletham over 7 years ago
Release of version 0.1