API: implement full sample simulation as experimental feature #297
+213
−33
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mgelsm
approved these changes
Sep 7, 2021
j-ittner
added a commit
that referenced
this pull request
Sep 8, 2021
* BUILD: pin Jinja2 version to <3.0 to prevent breaking the sphinx build (#286) * BUILD: pin Jinja2 version to <3.0 to prevent breaking the sphinx build * BUILD: update package requirements * DEV: update conda development environment in environment.yml * Update version number to 1.2.0 * Update version to 1.0.4 * DEV: pin scipy to 1.5 in environment.yml (#287) * API: remove FACET 1.0 legacy inspections (#288) * BUILD: pin jinja2 to ~=2.11 to prevent incompatibility with sphinx (#289) * BUILD: standardize azure pipeline across BCG Gamma repos (#292) * BUILD: create individual build tasks per package in single/matrix builds * DOC: remove comments * BUILD: use stable versions of pytools and sklearndf in matrix builds * FIX: revert path to /dist in CopyFiles task * BUILD: add more packages to conda build recipe * BUILD: use python 3.8 in conda host environment * BUILD: update build requirements incl. increased min requirements detail * BUILD: remove dev/* trigger * BUILD: downgrade typing_inspect to 0.6 in max build * BUILD: change numpy requirements syntax; improve upper version bounds * BUILD: add ipython to min/max matrix test dependencies * BUILD: use pip version syntax in pyproject.toml * BUILD: remove indirect package dependencies from default dependencies * BUILD: update pyproject.toml * BUILD: relax max build requirement for scipy * BUILD: use global variables for project and package name * BUILD: update standardized azure-pipelines.yml * FIX: re-remove _balancer.py * BUILD: fix pytools minimum versions in pyproject.toml * FIX: joblib version in min matrix build * BUILD: disable nightly builds on 1.1.x branch * BUILD: tweaks for 1.2.0 release (#295) * TEST: fix subsample to remove version dependence on random generator * FIX: de-duplicate sample index before intersecting with test splits * API: remove deprecated mirror package facet.simulation.partition * FIX: update nbsphinx to ~=0.8.5 to address new issue with jinja 3.0.0 * BUILD: fix max matrix versions at maximum * FIX: correct shap version spec in pyproject.toml * DOC: update RELEASE_NOTES.rst * BUILD: require pytools ~=1.2 and sklearndf ~= 1.2 * BUILD: require numpy 1.17 * BUILD: remove dev/* from azure pipeline triggers * FIX: rephrase max numpy requirement due to possible bug in conda-build * FIX: build sklearndf 1.2.x in azure-pipelines.yml * BUILD: use python 3.8 in host requirements * BUILD: remove unneeded trailing .* from pyproject.toml * BUILD: add missing version requirements to conda run config * BUILD: set matplotlib to 3.4 in max test * BUILD: change matplotlib==3 back to matplotlib==3.* for tox * BUILD: add inherited requirements so we can use them for testing * FIX: change boruta into boruta_py * FIX: change boruta_py back into boruta for tox builds (but not conda) * BUILD: make min requirements more specific to speed up build time * BUILD: remove conda dependencies with boruta Co-authored-by: Jörg Schneider <46053259+joerg-schneider@users.noreply.github.com> * BUILD: relax upper bounds of package requirements (#296) * BUILD: relax upper bounds of package requirements * BUILD: replace matplotlib with matplotlib-base * BUILD: use matplotlib-base only for conda * API: improve parameter checking for StratifiedBootstrapCV * TEST: add unit test for StratifiedBootstrapCV * FIX: don't require arg y in non-stratified bootstraps * DEV: update package requirements for facet-develop conda environment * API: implement full sample simulation as experimental feature (#297) * API: add hidden option to simulate single models on the full sample * DOC: fix an outdated (but hidden) docstring * TEST: validate partition frequencies * FIX: correct sort order of imports * Update __init__.py * FIX: use lists not tuples to create simulation results data frame Co-authored-by: Jörg Schneider <46053259+joerg-schneider@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds a hidden option to all
…Simulatorclasses that enables an alternative approach to calculating simulations and confidence intervals:LearnerCrossfit, run simulations using only one single model fit on the full dataset, and using the full dataset for all simulationsTo activate the alternative approach, set the hidden
full_sampleparameter toTrueafter creating a simulator: