microsoft
/
EconML
зеркало из https://github.com/py-why/EconML.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность
511 коммитов 42 Ветки 33 Теги 143 MiB
Граф коммитов

511 Коммитов

Автор SHA1 Сообщение Дата
Keith Battocchi 7462b66802 Add method for retrieving typical treatment values 2021-08-06 00:34:45 -04:00
Keith Battocchi 37c46fffb9 Improve error messages 2021-08-06 00:34:45 -04:00
Keith Battocchi 3ad50dafa6 Enable version bumps without rerunning all tests 2021-08-06 00:34:45 -04:00
Keith Battocchi 9c23ac97ca Prepare v0.12.0b5 release 2021-08-03 14:24:44 -04:00
Keith Battocchi 4e6172f962 Update EconML logo 2021-08-03 14:24:44 -04:00
vsyrgkanis 821be59323
Changed the meaning of effect of treatment in individualized policy (#504) 
* made individualized policy return the delta from the current treatment as effect of recommendation, as compared to delta with respect to baseline. this seems more interpretable.

* linting errors. Also added confidence intervals to the heterogeneity tree output dictionary
 2021-08-02 16:37:04 -04:00
Maggie Hei 8b0c33aaf5
update notebook for causal analysis (#499) 
* update notebook for causal analysis
 2021-07-21 18:22:25 -04:00
Keith Battocchi eade2b1071 Prepare v0.12.0b4 release 2021-07-10 14:57:21 -04:00
Keith Battocchi c3d5bd635f Add additional causal analysis flags 2021-07-10 14:57:21 -04:00
vsyrgkanis 0cec5ab3d4
Update README.md (#491) 
changing citation
 2021-06-30 16:26:24 -04:00
Snow Storm 452beb2025 Fixing spelling 
Description of changes: Fixing spelling on notebook 

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
 2021-06-30 16:08:35 -04:00
Keith Battocchi 9bb06f9279 Prepare v0.12.0b3 release 2021-06-26 12:20:02 -04:00
Keith Battocchi b4d5716fbd Switch causal analysis policy tree output format 2021-06-26 12:20:02 -04:00
Keith Battocchi b1a7f44005 Fix policy computation with array outcome 2021-06-26 12:20:02 -04:00
Keith Battocchi d493d285c2 Ensure tranforming InferenceResults maintains shape 2021-06-26 12:20:02 -04:00
Chris Gregory 4551a84ad3
Make graphviz a soft dependency (#486) 
Make graphviz a soft visualization dependency
 2021-06-26 03:49:27 +00:00
Chris Gregory b4e72bc304
Upgrade shap dependency (#487) 
Upgrade shap dependency to ~0.39.0
 2021-06-26 00:15:40 +00:00
Chris Gregory d63f8b260b Include cardinality in categorical expansion warning 2021-06-25 16:05:44 -04:00
Keith Battocchi d5639db2eb Prepare v0.12.0b2 release 2021-06-19 08:55:30 -04:00
Keith Battocchi ffd76d16c6 Keep track of failed features instead of throwing 2021-06-19 08:55:30 -04:00
Keith Battocchi af10eca2dc Prepare v0.12.0b1 release 2021-06-08 00:11:48 -04:00
Keith Battocchi a1e0f3e483 Enable X=None when interpreting policy trees 2021-06-08 00:11:48 -04:00
Keith Battocchi 50418ccb86 Raise ForestDML clip bound to stabilize inference 2021-06-08 00:11:48 -04:00
Keith Battocchi 46268cea56 Improve causal analysis functionality 2021-06-08 00:11:48 -04:00
Maggie Hei 420dfceba9
Add regression notebook under Solutions (#474) 
* add regression notebook
 2021-06-02 14:52:30 -04:00
Keith Battocchi dbbd948871 Prepare 0.11.1 release 2021-05-19 02:12:32 -04:00
Keith Battocchi 64c4d953ee Add treatment recommendations to causal analysis 2021-05-19 02:12:32 -04:00
Keith Battocchi aa56e7d025 Fix ortho_learner docs 2021-05-19 02:12:32 -04:00
Keith Battocchi 3b9375e120 Correct installation and testing instructions 2021-05-19 02:12:32 -04:00
Keith Battocchi 20eb438b03 Correct statsmodels version bound 2021-05-19 02:12:32 -04:00
Keith Battocchi c6bdef2c75 Add init args to causal insights dictionary 2021-05-19 02:12:32 -04:00
Keith Battocchi 102b461244 Enable categorical range checking 2021-05-19 02:12:32 -04:00
Keith Battocchi b0b373b1f7 Fix ortholearner and causal analysis pickling 2021-05-19 02:12:32 -04:00
Maggie Hei 149a17d7f4
Add causal analysis notebook (#466) 
* add causal analysis notebook
 2021-05-17 17:55:17 -04:00
vsyrgkanis 5ab177ffd1
fixed max_samples=.5 in default grid for cfddml.tune (#462) 
* fixed max_samples=.5 in default grid for cfddml.tune

* relaxed rscorer test

* fixed rounding versus flooring in clauclating size of subsample. Fixed tests in grf_python so that this error would have been caught.
 2021-05-11 07:45:37 -04:00
Keith Battocchi 472dd8e964 Prepare v0.11.0 release 2021-05-08 21:24:43 -04:00
Keith Battocchi 83b9e1471f Add causal analysis 2021-05-08 21:24:43 -04:00
Keith Battocchi 70f7d0c9ae Enable CausalForestDML to work with pandas 2021-05-08 21:24:43 -04:00
Chris Gregory 4a409cd6d8
Fix tree exporter utilities import (#459) 
Fix tree exporter utilities import
 2021-05-06 21:01:33 +00:00
Maggie Hei b53c0e3cd1
set check_additivity=False for CausalForestDML (#458) 
* set check_additivity=False for TreeExplainer
 2021-05-05 20:40:08 -04:00
Maggie Hei 7e74ed8fb6
add lalonde notebook (#448) 
add lalonde notebook
 2021-05-04 13:51:54 -04:00
vsyrgkanis 5bf448a02a
Vasilis/casestudy (#441) 
* changed how aggregation happens in policy ensembles. Added policy learning to mulit-investment attribution. Fixed some mistkaes in that case study

* added clipping to the denominator in the dr correction to avoid division by zero, in the dr estimate of ate and att in cfdml.
 2021-05-03 11:03:56 -04:00
Maggie Hei 9630421315
create both sample_weight and freq_weight for fit (#439) 
* Make StatsmodelsLinearRegresssion allow fractional weight for each individual sample, which is sample_weight. Rename the weights used as the count of observations for corresponding sample_var as freq_weight.
* For all the child classes inherit from _OrthoLearner, only expose freq_weight and sample_weight when the final stage is StatsmodelsLinearRegression or its parent class.
* Fix a few places to make sure it works consistently with un-summarized data when both weights exist.
 2021-03-31 12:56:39 -04:00
vsyrgkanis 056fb30b49
addeed policy learning to readme (#440) 
* addeed policy learning to readme
 2021-03-23 13:05:28 -04:00
Keith Battocchi 4572bdab3e Prepare v0.10.0 release 2021-03-22 19:57:36 -04:00
vsyrgkanis 98b2bf3234
Vasilis/policy (#377) 
* added policy learning module

* added cython policy tree and policy forest

* extended policy cate interpreter to interpret multiple treatments using the new policy tree

* added doubly robust policy learning tree and doubly robust policy learning forest

* fixed randomness in weightedkfold, that was causing tests to fail due to non-fixed-randomness behavior

* added notebook on policy learning
 2021-03-22 18:56:25 -04:00
Maggie Hei cac4c3e757
Fix deprecated args and dowhy wrapper (#434) 
* remove the deprecated args in ForestDRLearner and add strong condition on search params from dowhy wrapper
* remove all the deprecated n_splits/n_crossfit_splits
 2021-03-22 11:46:54 -04:00
Maggie Hei 7005b24253
Store nuisance models and scores (#433) 
* store nuisance models and scores for each mc iteration and adapt the score function
 2021-03-19 16:52:28 -04:00
Keith Battocchi fed1a9a9fc Bump version to 0.9.2 2021-03-11 19:51:18 -05:00
vsyrgkanis 41ffb03a91
Vasilis/drate (#391) 
* enalbed training data doubly robust ate and att calculation and inference. added a summary functionality to the causal forest.

* added ate__inference in dml notebook

* removed warm start from cfdml.
 2021-03-11 17:06:01 -05:00