Releases: alkaline-ml/pmdarima
Version 1.1.1
v1.1.1 is a patch release in response to #104
-
Deprecates the
ARIMA.add_new_observations
method. This method originally was designed to support updating the endogenous/exogenous arrays with new observations without changing the model parameters, but achieving this behavior for each of statsmodels'ARMA
,ARIMA
andSARIMAX
classes proved nearly impossible, given the extremely complex internals of statmodels estimators. -
Replace
ARIMA.add_new_observations
withARIMA.update
. This allows the user to update the model with new observations by takingmaxiter
new steps from the existing model coefficients and allowing the MLE to converge to an updated set of model parameters. -
Change default
maxiter
to None, using 50 for seasonal models and 500 for non-seasonal models (as statsmodels does). The default value used to be 50 for all models. -
New behavior in
ARIMA.fit
allowsstart_params
andmaxiter
to be passed as**fit_args
, overriding the use of their corresponding instance attributes.
Version 1.1.0
Release 1.1.0 adds:
ARIMA.plot_diagnostics
method, as requested in #49- Adds new arg to
ARIMA
constructor andauto_arima
:with_intercept
(default is True). - New default for
trend
is no longer'c'
, it isNone
. - Added
to_dict
method toARIMA
class to address #54 - The
'PMDARIMA_CACHE'
and'PMDARIMA_CACHE_WARN_SIZE'
environment variables are
now deprecated, since they no longer need to be used. They will be removed in v1.2.0 - Added versioned documentation. All releases' doc (from 0.9.0 onward) is now available
atalkaline-ml.com/pmdarima/<version>
- Python 3.7 support(!!)
Version 1.0.0
v1.0.0
-
Wheels will no longer be built for Python versions < 3.5. You may still be able to build
from source, but support for 2.x python versions will diminish in future versions. -
Migrate namespace from 'pyramid-arima' to 'pmdarima'. This is due to the fact that
a growing web-framework (also named Pyramid) is causing namespace collisions when
both packages are installed on a machine. See Issue #34
for more detail. -
Remove redundant Travis tests
-
Automate documentation build on Circle CI
-
Move lots of the build/test functionality into the
Makefile
for ease. -
Warn for impending deprecation of various environment variable name changes. The following
will be completely switched over in version 1.2.0:'PYRAMID_MPL_DEBUG'
will become'PMDARIMA_MPL_DEBUG'
'PYRAMID_MPL_BACKEND'
will become'PMDARIMA_MPL_BACKEND'
'PYRAMID_ARIMA_CACHE_WARN_SIZE'
will become'PMDARIMA_CACHE_WARN_SIZE'
Version 0.9.0
v0.9.0
Bug fixes:
-
Explicitly catch case in
auto_arima
where a value ofm
that is too large may over-estimate
D
, causing the time series to be differenced down to an empty array. This is now handled by
raising a separate error for this case that better explains what happened. -
Re-pickling an
ARIMA
will no longer remove the location on disk of the cachedstatsmodels
ARIMA models. Older versions encountered an issue where an older version of the model would be
reinstated and immediately fail due to an OSError since the cached state no longer existed. This
means that a user must be very intentional about clearing out the pyramid cache over time (addresses #39) -
Added pyramid cache check on initial import to warn user if the cache size has grown too large.
-
If
d
orD
are explicitly defined forauto_arima
(rather thanNone
), do not
raise an error if they exceedmax_d
ormax_D
, respectively (sort of addresses #37)
New features:
-
Added Circle CI for validating PyPy builds (rather than CPython)
-
Deploy python wheel for version 3.6 on Linux and Windows (addresses #38)
Forward looking
- Include warning for upcoming package name change (
pmdarima
).
Version 0.8.1
v0.8.1
-
ARIMA
instance attributes- The
pkg_version_
attribute (assigned on modelfit
) is new as of version 0.8.1.
On unpickling, if the current Pyramid version does not match the version under which it
was serialized, aUserWarning
will be raised.
- The
-
Addition of the
_config.py
file at the top-level of the package- Specifies the location of the ARIMA result pickles
- Specifies the ARIMA result pickle name pattern
-
Fix bug (Issue #30) in
ARIMA
where using CV with differencing and no seasonality caused a dim mismatch in the model's
exog array and its endog array -
New dataset: woolyrnq (from R's
forecast
package). -
Visualization utilities available at the top level of the package:
plot_acf
plot_pacf
autocorr_plot
-
Updated documentation with significantly more examples and API references.
Version 0.7.1
Bug fixes:
- Out-of-sample scoring/fitting is now handled appropriately (see Issue #28).
- Fixed
TypeError
onARMA
predictions whered=0
(see Issue #29).
New features:
- Created a
add_new_samples
method forARIMA
models so newly observed samples can be added to a model without updating its parameters. This allows future forecasts to consider the latest ground-truth samples. - Added confidence intervals to predictions. See Issue #20
Documentation:
- Overhauled documentation with new formatting
- Migrated off of readthedocs and onto
gh-pages
Version 0.6.5
Much faster ARIMA computation since we don't compute the U or V in the SVD component.
Version 0.6.2
Version 0.6.2 is a hot fix that addresses Issue #16, where Pyramid could not be installed via PIP on an EC2 Linux machine.
Version 0.5.1
This release adds a hotfix to the v0.5 release, which includes the requirements.txt
in the MANIFEST.in
and allows pip installs from pypi.
Version 0.5
v0.5 Bump version