Version 2.1.7

• Limit allowed versions of pandas <2.1 to avoid a bug
• Update versions of indirect dependencies in requirements.txt and notebook_requirements.txt

For full details please see the change log

Version 2.1.6

• Fix NonExistentTimeError in rdtools.clearsky_temperature.get_clearsky_tamb()
• rdtools.degradation.degradation_classical_decomposition() now executes significantly faster
• Increased the minimum versions of several dependencies
  • pandas increased to 1.3.0 (released July 2, 2021)
  • numpy to 1.17.3 (released October 17, 2019)
  • statsmodels to 0.11.0 (released February 21, 2020)
  • scipy to 1.2.0 (released December 17, 2018)

For full details, please see the change log

Version 2.1.5

Adds support for Pandas 2.0 (see #362)

Version 2.2.0-beta.1

Adds a framework for applying filters to aggregated data in TrendAnalysis. For full details see the change log.

Version 2.1.4

This patch includes two bug fixes:

• Fixes an error from degradation_year_on_year() for datasets that are exactly 2 years long
• soiling_interval_plot() no longer ignores some optional parameters.

For full details see the change log

Version 2.2.0-beta.0

This release includes several new enhancements:

• Detailed degradation summary plot
• Degradation time series plot for assessing non-linear degradation
• The Combined estimation OF Degradation and Soiling (CODS) algorithm implemented in the soiling module.

For full details, please see the change log

Version 2.1.3

What's Changed

• Drop python 3.6
• Add python 3.10
• Fixed a plotting issue in plotting.availability_summary_plots() with newer matplotlib versions, as well as an axis labeling error
• Full Change log

All Changes: 2.1.2...2.1.3

Version 2.1.2

What's Changed

• Change XGBoost model to version-stable json
• Change recommended citation
• For details see the change log

All Changes: 2.1.1...2.1.2

Version 2.1.1

This patch release temporarily requires xgboost < 1.5.0. A future update will allow a wider version range. (#301, #297 )

Version 2.1.0

This update includes changes to the experimental soiling functions, as well as a host of added functionality including the full release of an object oriented end-to-end analysis API and new algorithms for clipping detection and filtering. See the changelog for full details.