Bibliographic record
Abstract
Contributors to this version: Trevor James Smith (@Zeitsperre), Juliette Lavoie (@juliettelavoie), Éric Dupuis (@coxipi), Pascal Bourgault (@aulemahal). Announcements xclim now supports testing against tagged versions of Ouranosinc/xclim-testdata _ in order to support older versions of xclim. For more information, see the Contributing Guide for more details. (PR/1339). xclim v0.42.0 will be the last version to explicitly support Python3.8. (GH/1268, PR/1344). New features and enhancements Two previously private functions for selecting a day of year in a time series when performing calendar conversions are now exposed. (GH/1305, PR/1317). New functions are: xclim.core.calendar.yearly_interpolated_doy xclim.core.calendar.yearly_random_doy scipy is no longer pinned below v1.9 and lmoments3>=1.0.5 is now a core dependency and installed by default with pip. (GH/1142, PR/1171). Fix bug on number of bins in xclim.sdba.propeties.spatial_correlogram. (PR/1336) Add resample_before_rl argument to control when resampling happens in maximum_consecutive_{frost|frost_free|dry|tx}_days and in heat indices (in _threshold) (GH/1329, PR/1331) Add xclim.ensembles.make_criteria to help create inputs for the ensemble-reduction methods. (GH/1338, PR/1341). Bug fixes Warnings emitted from regular usage of some indices (snowfall_approximation with method="brown", effective_growing_degree_days) due to successive convert_units_to calls within their logic have been silenced. (PR/1319). Fixed a bug that prevented the use of the sdba_encode_cf option with xarray 2023.3.0 (PR/1333). Fixed bugs in xclim.core.missing and xclim.sdba.base.Grouper when using pandas 2.0. (PR/1344). Breaking changes The call signatures for xclim.ensembles.create_ensemble and xclim.ensembles._base._ens_align_dataset have been deprecated. Calls to these functions made with the original signature will emit warnings. Changes will become breaking in xclim>=0.43.0.(GH/1305, PR/1317). Affected variable: mf_flag (bool) -> multifile (bool) The indice and indicator for last_spring_frost has been modified to use tasmin by default, reflecting its docstring and literature definition (GH/1324, PR/1325). following indices now accept the op argument for modifying the threshold comparison operator (PR/1325): snw_season_length, snd_season_length, growing_season_length, frost_season_length, frost_free_season_length, rprcptot, daily_pr_intensity In order to support older environments, pandas is now conditionally pinned below v2.0 when installing xclim on systems running Python3.8. (PR/1344). Bug fixes xclim.indices.run_length.last_run nows works when freq is not None. (GH/1321, PR/1323). Internal changes Added xclim to the ouranos Zenodo community . (PR/1313). Significant documentation adjustments. (GH/1305, PR/1308): The CONTRIBUTING page has been moved to the top level of the repository. Information concerning the licensing of xclim is clearly indicated in README. sphinx-autodoc-typehints is now used to simplify call signatures generated in documentation. The SDBA module API is now found with the rest of the User API documentation. HISTORY.rst has been renamed CHANGES.rst, to follow dask-like conventions. Hyperlink targets for individual indices and indicators now point to their entries under API or Indices. Module-level docstrings have migrated from the library scripts directly into the documentation RestructuredText files. The documentation now includes a page explaining the reasons for developing xclim and a section briefly detailing similar and related projects. Markdown explanations in some Jupyter Notebooks have been edited for clarity Removed Mapping abstract base class types in call signatures (dict variables were always expected). (PR/1308). Changes in testing setup now prevent test_mean_radiant_temperature from sometimes causing a segmentation fault. (GH/1303, PR/1315). Addressed a formatting bug that caused Indicators with multiple variables returned to not be properly formatted in the documentation. (GH/1305, PR/1317). tox now include sbck and eofs flags for easier testing of dependencies. CI builds now test against sbck-python @ master. (PR/1328). upstream CI tests are now run on push to master, at midnight, and can also be triggered via workflow_dispatch. Failures from upstream build will open issues using xarray-contrib/issue-from-pytest-log. (PR/1327). Warnings from set _version_deprecated within Indicators now emit FutureWarning instead of DeprecationWarning for greater visibility. (PR/1319). The Graphics section of the Usage notebook has been expanded upon while grammar and spelling mistakes within the notebook-generated documentation have been reduced. (GH/1335, PR/1338, suggested from PyOpenSci Software Review). The Contributing guide now lists three separate subsections to help users understand the gains from optional dependencies. (GH/1335, PR/1338, suggested from PyOpenSci Software Review).
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.017 |
| Meta-epidemiology (narrow) | 0.003 | 0.003 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.006 |
| Open science | 0.006 | 0.007 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.295 | 0.394 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".