Bibliographic record
Abstract
History v0.6.0 (2023-05-04) Contributors to this version: Trevor James Smith (@Zeitsperre), Juliette Lavoie (@juliettelavoie), Pascal Bourgault (@aulemahal), Gabriel Rondeau-Genesse (@RondeauG). Announcements xscen is now offered as a conda package available through Anaconda.org. Refer to the installation documentation for more information. (GH/149, PR/171). Deprecation: Release 0.6.0 of xscen will be the last version to support xscen.extract.clisops_subset. Use xscen.spatial.subset instead. (PR/182, PR/184). Deprecation: The argument region, used in multiple functions, has been slightly reformatted. Release 0.6.0 of xscen will be the last version to support the old format. (GH/99, GH/101, PR/184). ### New features and enhancements New 'cos-lat' averaging in spatial_mean. (GH/94, PR/125). Support for computing anomalies in compute_deltas. (PR/165). Add function diagnostics.measures_improvement_2d. (PR/167). Add function regrid.create_bounds_rotated_pole and automatic use in regrid_dataset and spatial_mean. This is temporary, while we wait for a functionning method in cf_xarray. (PR/174, GH/96). Add spatial submodule with functions creep_weights and creep_fill for filling NaNs using neighbours. (PR/174). Allow passing GeoDataFrame instances in spatial_mean's region argument, not only geospatial file paths. (PR/174). Allow searching for periods in catalog.search. (GH/123, PR/170). Allow searching and extracting multiple frequencies for a given variable. (GH/168, PR/170). New masking feature in extract_dataset. (GH/180, PR/182). New function xs.spatial.subset to replace xs.extract.clisops_subset and add method "sel". (GH/180, PR/182). Add long_name attribute to diagnostics. ( PR/189). Added a new YAML-centric notebook (GH/8, PR/191). New utils.standardize_periods to standardize that argument across multiple functions. (GH/87, PR/192). New coverage_kwargs argument added to search_data_catalogs to allow modifying the default values of subset_file_coverage. (GH/87, PR/192). ### Breaking changes 'mean' averaging has been deprecated in spatial_mean. (PR/125). 'interp_coord' has been renamed to 'interp_centroid' in spatial_mean. (PR/125). The 'datasets' dimension of the output of diagnostics.measures_heatmap is renamed 'realization'. (PR/167). _subset_file_coverage was renamed subset_file_coverage and moved to catalog.py to prevent circular imports. (PR/170). extract_dataset doesn't fail when a variable is in the dataset, but not variables_and_freqs. (PR/185). The argument period, used in multiple function, is now always a single list, while periods is more flexible. (GH/87, PR/192). The parameters reference_period and simulation_period of xscen.train and xscen.adjust were renamed period/periods to respect the point above. (GH/87, PR/192). ### Bug fixes Forbid pandas v1.5.3 in the environment files, as the linux conda build breaks the data catalog parser. (GH/161, PR/162). Only return requested variables when using DataCatalog.to_dataset. (PR/163). compute_indicators no longer crashes if less than 3 timesteps are produced. (PR/125). xarray is temporarily pinned below v2023.3.0 due to an API-breaking change. (GH/175, PR/173). `xscen.utils.unstack_fill_nan`` can now handle datasets that have non dimension coordinates. (GH/156, PR/175). extract_dataset now skips a simulation way earlier if the frequency doesn't match. (PR/170). extract_dataset now correctly tries to extract in reverse timedelta order. (PR/170). compute_deltas no longer creates all NaN values if the input dataset is in a non-standard calendar. (PR/188). ### Internal changes xscen now manages packaging for PyPi and TestPyPI via GitHub workflows. (PR/159). Pre-load coordinates in extract.clisops_subset (PR/163). Minimal documentation for templates. (PR/163). xscen is now indexed in Zenodo, under the ouranos community of projects. (PR/164). Added a few relevant Shields to the README.rst. (PR/164). Better warning messages in _subset_file_coverage when coverage is insufficient. (PR/125). The top-level Makefile now includes a linkcheck recipe, and the ReadTheDocs configuration no longer reinstalls the llvmlite compiler library. (PR/173). The checkups on coverage and duplicates can now be skipped in subset_file_coverage. (PR/170). Changed the ProjectCatalog docstrings to make it more obvious that it needs to be created empty. (GH/99, PR/184). Added parse_config to creep_fill, creep_weights, and reduce_ensemble (PR/191).
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 | 0.018 |
| Meta-epidemiology (narrow) | 0.004 | 0.004 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.004 | 0.004 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.010 | 0.009 |
| Open science | 0.007 | 0.007 |
| Research integrity | 0.003 | 0.005 |
| Insufficient payload (model declined to judge) | 0.665 | 0.813 |
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".