eluc_1.0.6_baylearn

App update for BayLearn along with some other experimental changes and bugfixes.

What's Changed

• Open source prescriptor training using PyTorch implementation of NSGA-II by @danyoungday in #67
• Move heuristics into their own class by @danyoungday in #68
• Updated miscellaneous documentation and readme by @danyoungday in #70
• Upgraded seed creating to handle argument passing and perform validation by @danyoungday in #69
• Rudimentary Requirements File by @danyoungday in #76
• Create yaml file for ELUC use case CI by @danyoungday in #77
• Updated training script for torch prescriptors to allow config by @danyoungday in #75
• Fix git merge issues with seeding by @danyoungday in #74
• Updated paper link in readme to point to risto's site instead of arxiv by @danyoungday in #73
• More linting. Ignore ESP file. Clean redundancy in CI. by @danyoungday in #78
• Updated readme with better instructions by @danyoungday in #80
• Refactor predictor loading and saving by @danyoungday in #81
• Fully trained torch prescriptors and updated experiments by @danyoungday in #72
• Redo experiments with fixed dataset by @danyoungday in #71
• Added ability to save and load from huggingfce by @danyoungday in #83
• Fix distance calcuation in NSGA-II by @danyoungday in #86
• Refactor app with new predictor/prescriptor by @danyoungday in #87
• Refactor Prescriptors and project to handle new prescription by @danyoungday in #89
• Clean up app by moving logic into "Component" files from app.py by @danyoungday in #91
• Refactor prediction and prescription by @danyoungday in #93
• Fix domination by @danyoungday in #94
• Predictor Contribution by @danyoungday in #96
• Added Flake8 to github actions and cleaned code to pass it by @danyoungday in #97
• Trimmed requirements down and fixed issue with bad download by @danyoungday in #99
• Added tensorboard back into requirements by @danyoungday in #100
• Fixed indentation error in github actions for flake by @danyoungday in #98
• Import abstract classes from SDK by @danyoungday in #101
• Move predictors out and into SDK, remove storing of CAO in models by @danyoungday in #103
• got presc experiments to work with new presc manager by @danyoungday in #95
• Filter app by @danyoungday in #104
• Fixed issue scaling ELUC slider with change slider by @danyoungday in #105

Full Changelog: eluc_1.0.5_gecco...eluc_1.0.6_baylearn

eluc_1.0.5_gecco

Further revisions to the project to reflect work done on the paper. Predictors were moved out of the notebook and into their own classes. The app was refactored to work with the new changes. An error was fixed where the first test year was bleeding into the train set.

What's Changed

• Demo by @danyoungday in #31
• Merge dash demo into main by @danyoungday in #32
• Restructured demo app and made it suitable for deployment by @danyoungday in #33
• Improved models with linear regression and added every country to map by @danyoungday in #34
• Minor tweaks to get app deployable by @danyoungday in #35
• Updated prescriptor to not require unileaf-util DataEncoder. by @danyoungday in #37
• Added tests for prescriptor and tweaked prescriptor by @danyoungday in #44
• Add test for no change to ELUC by @jacobbieker in #30
• Refactored project + some bug fixes by @danyoungday in #56
• Add a README to for the ELUC use case by @ofrancon in #57
• Added links to repo and paper by @danyoungday in #58
• Added data processing files for ELUC data upload to huggingface or download by @danyoungday in #60
• Refactor predictors by @danyoungday in #61
• Base predictor implementation: NN, LinReg, RF by @danyoungday in #62
• Added custom encoder so that we no longer reference esp by @danyoungday in #64
• Added predictor experiments from paper by @danyoungday in #63
• Transferred prescriptor code to repo by @danyoungday in #65
• Updated eluc data to not duplicate the year 2012 by @danyoungday in #66

Full Changelog: eluc_1.0.4_neurips...eluc_1.0.5_gecco

eluc_1.0.4_neurips

This version marks the completed experiments that were used in the version of the Discovering Effective Policies for Land-Use Planning paper which was a spotlight presentation at the NeurIPS 2023 Workshop: Tackling Climate Change with Machine Learning. Experiments are run in use_cases/eluc/model/project_LandUseChange_experiment_forest_run_1184.ipynb.

Full Changelog: eluc_1.0.3...eluc_1.0.4_neurips

eluc_1.0.3

Removed LEAF dependency and fixed bugged prescriptors.

What's Changed

• Updated prescriptor to not require unileaf-util DataEncoder. by @danyoungday in #37
• Added tests for prescriptor and tweaked prescriptor by @danyoungday in #44
• Add test for no change to ELUC by @jacobbieker in #30

Full Changelog: eluc_1.0.2...eluc_1.0.3

eluc_1.0.2

Minuscule tweaks to allow deployment.
Updated typo in the README and increased timeout time in gunicorn to prevent timeout error.

What's Changed

• Minor tweaks to get app deployable by @danyoungday in #35

Full Changelog: eluc_1.0.1...eluc_1.0.2

eluc_1.0.1

Added every country to map.
Changed prescriptors to be evaluated on linear regression model trained on West Europe.
Added some predictors specific to region.
Added a cute favicon instead of the default dash one!

What's Changed

• Improved models with linear regression and added every country to map by @danyoungday in #34

Full Changelog: eluc_1.0.0...eluc_1.0.1

eluc_1.0.0

Emissions from Land Use Change version 1.0.0