Change Log¶
v0.5.3 (2017-06-26)¶
Implemented enhancements:
- Weighted KPIs is only implemented in regular delta #114
Fixed bugs:
- Assumption of nan when computing weighted KPIs #119
- Weighted KPIs is only implemented in regular delta #114
- Percentiles value is lost during computing group_sequential_delta #108
Closed issues:
- Failing early stopping unit tests #85
Merged pull requests:
- OCTO-1804: Optimize the loading of .stan model in expan. #126 (daryadedik)
- Test travis python version #125 (shansfolder)
- OCTO-1619 Cleanup ExpAn code #124 (shansfolder)
- OCTO-1748: Make number of iterations as a method argument in _bayes_sampling #123 (daryadedik)
- OCTO-1615 Use Python builtin logging instead of our own debugging.py #122 (shansfolder)
- OCTO-1711 Support weighted KPIs in early stopping #121 (shansfolder)
- Fixed a few bugs #120 (shansfolder)
- OCTO-1614 cleanup module structure #115 (shansfolder)
- OCTO-1677 : fix missing .stan files #113 (gbordyugov)
- Bump version 0.5.1 -> 0.5.2 #112 (mkolarek)
v0.5.2 (2017-05-11)¶
Merged pull requests:
- OCTO-1502 support **kwargs for four delta functions #111 (shansfolder)
- new version 0.5.1 #107 (mkolarek)
v0.5.1 (2017-04-20)¶
Implemented enhancements:
- Derived KPIs are passed to Experiment.fixed_horizon_delta() but never used in there #96
Merged pull requests:
- OCTO-1501: bugfix in Results.to_json() #105 (gbordyugov)
- OCTO-1502 removed variant_subset parameter... #104 (gbordyugov)
- OCTO-1540 cleanup handling of derived kpis #102 (shansfolder)
- OCTO-1540: cleanup of derived kpi handling in Experiment.delta() and … #97 (gbordyugov)
- Merge dev to master for v0.5.0 #94 (mkolarek)
v0.5.0 (2017-04-05)¶
Implemented enhancements:
Fixed bugs:
- Experiment.sga() assumes features and KPIs are merged in self.metrics #87
- pctile can be undefined in
Results.to\_json\(\)
#78
Closed issues:
- Results.to_json() => TypeError: Object of type ‘UserWarning’ is not JSON serializable #77
- Rethink Results structure #66
Merged pull requests:
- new dataframe tree traverser in to_json() #92 (gbordyugov)
- updated requirements.txt to have ‘greater than’ dependencies instead … #89 (mkolarek)
- pip version requirement #88 (gbordyugov)
- Test #86 (s4826)
- merging in categorical binning #84 (gbordyugov)
v0.4.4 (2017-02-09)¶
Implemented enhancements:
- Add argument assume_normal and treatment_cost to calculate_prob_uplift_over_zero() and prob_uplift_over_zero_single_metric() #26
- host intro slides (from the ipython notebook) somewhere for public viewing #10
Closed issues:
- migrate issues from github enterprise #20
v0.4.2 (2016-12-08)¶
Fixed bugs:
- frequency table in the chi square test doesn’t respect the order of categories #56
v0.4.0 (2016-08-19)¶
Closed issues:
- Support ‘overall ratio’ metrics (e.g. conversion rate/return rate) as opposed to per-entity ratios #44
v0.2.5 (2016-05-30)¶
Implemented enhancements:
- Implement __version__ #14
Closed issues:
- upload full documentation! #1