Bibliographic record
Abstract
Release Notes This release includes a number of substantial changes to NiMARE. Major changes We've added PyMARE as a dependency! PyMARE is a general-purpose meta-analysis library in Python that we now use to perform our image-based meta-analyses. For image-based meta-analyses, we also now have a transforms module to calculate new image types from available data. Datasets now have a number of attributes retained as properties, which will break compatibility with Datasets from older versions of NiMARE. We now have multiple methods for converting summary statistics (e.g., ALE, OF) to p-values in all of our major CBMA algorithms, thanks to @tyarkoni! The two current methods for each algorithm are a fast, but slightly less accurate, "analytic" method and a slower, but more accurate, "empirical" method. For ALE, We generally recommend the "analytic" method for maximum compatibility with GingerALE. The implementations of these algorithms have also been streamlined and sped up somewhat. We have a new generate module for simulating coordinate-based datasets, thanks to @jdkent! A number of modules, classes, and functions that were not yet implemented have been pruned from the API to make it easier to work with. Don't worry, we're still planning to get around to them at some point. Changes [FIX] Fix the warnings about mismatched kernels and estimators (#425) @tsalo [FIX] Add nullhist_to_p and crop invalid p-values (#409) @tsalo [TST] Do not download test peaks2maps to tmpdir (#419) @tsalo [FIX] Restructure Peaks2MapsKernel to operate like other kernels (#410) @tsalo [ENH] Improve convergence between ALE null methods (#411) @tsalo [DOC] Add warnings for CBMA kernel/estimator mismatch (#416) @tsalo [FIX] Remove rows with empty abstract before running LDAModel (#414) @JulioAPeraza [FIX] Sort all arrays and DataFrames in Dataset by ID (#402) @tsalo [FIX] Allow no coordinates in a dataset (#407) @jdkent [ENH] Add analytic null method to KDA estimator (#397) @tsalo [FIX] Use unzipped mask as temporary fix (#401) @tsalo [DOC] Update API and examples (#395) @tsalo [REF] CBMA re-organization and improvement (#393) @tyarkoni [MAINT] Pin to PyMARE 0.0.2 (#391) @tsalo [TST] Test both analytic and empirical methods in ALE and MKDA (#380) @jdkent [FIX] Change default seed to None (#392) @jdkent [PERF] Various performance improvements (#386) @tyarkoni Add performance tweaks to ALE analytical null generation (#390) @tyarkoni fix tests (#387) @tyarkoni [FIX] respect n_noise_foci value (#382) @jdkent [ENH] Add analytic null method to MKDADensity (#375) @tsalo [ENH] Add empirical null method to density-based CBMA Estimators (#372) @tsalo [REF] Refactor KernelTransformer hierarchy (#369) @tyarkoni [ENH] Add generate module (#343) @jdkent [FIX] enforce correct lowest p-value (#365) @jdkent [FIX] Treat vfwe as an array of floats for KDA (#362) @jdkent [DOC] Update roadmap.rst (#359) @tsalo [DOC] Add example of combining kernels and CBMA estimators (#346) @koudyk [MAINT] Add Dorota Jarecka to Zenodo file (#358) @djarecka [MAINT] Add Enrico Glerean's affiliation and ORCID (#357) @eglerean [ENH] Clip p-values based on number of permutations (#353) @tsalo [REF] Remove unused alpha argument in statsmodels call (#354) @tsalo [ENH] Replace TTest with PermutedOLS (#304) @tsalo [REF] Reduce dependencies (#345) @tsalo [ENH] Add Neurosynth data fetcher (#342) @tsalo [INFRA] Add json describing filename convention (#338) @tsalo [DOC] Enable CBMA example (#337) @tsalo [FIX] Add private setter method for Dataset.ids (#336) @tsalo [REF] More low-memory work (#334) @tsalo [FIX, DOC] Change natural log to base-ten and document output naming convention (#333) @tsalo [FIX] Pin setuptools again (#331) @tsalo [FIX] Update setuptools version (#330) @tsalo [FIX] Add setuptools to requirements (#329) @tsalo [TST] Add test for peaks2maps (#328) @tsalo [FIX, TST] Fix and test CorrelationDistributionDecoder (#327) @tsalo [TST] Use temporary directories with automatic teardown (#326) @tsalo [REF] Speed up CorrelationDecoder (#324) @tsalo [ENH] Support Dataset transformations in kernel transformers (#320) @tsalo [ENH] Add PairwiseCBMAEstimator class and add low_memory option to ALESubtraction (#319) @tsalo [TST] Improve meta-analysis tests (#318) @tsalo [DOC] Fix Lancaster xform and Sleuth conversion docstrings (#317) @tsalo [TST] Improve nimare.io test coverage (#314) @tsalo [REF] Reduce duplication by calling _check_ncores (#313) @tsalo [REF] Remove generate_cooccurrence (#312) @tsalo [REF] Operate on arrays in ALESubtraction (#311) @tsalo [TST] Add flake8-black to test requirements (#300) @akimbler [FIX] Support multiple header lines in Sleuth text files (#310) @tsalo [FIX] Operate on copy of df in extract_cogat() (#306) @tsalo [MAINT] Update setup configuration (#303) @tsalo [REF] Sort imports alphabetically (#299) @tsalo [REF] Run automated code formatting with black (#296) @tsalo [DOC] Remove whitespace from README (#295) @tsalo [MAINT, TST] Drop 3.5 support. Add tests for Python 3.7 and 3.8. (#293) @tsalo [MAINT] Delete unused files (#291) @tsalo [MAINT] Increase minimum tensorflow to 2.0.0 (#290) @tsalo [FIX] Update peaks2maps w.r.t. recent changes in the API (#287) @tsalo [FIX] Raise an error in Decoders if no features remain (#284) @tsalo [REF] Move CBMA methods up a level (#283) @tsalo [REF] Rename RandomEffectsGLM to TTest (#282) @tsalo [ENH] Split DerSimonianLaird and Hedges IBMA estimators (#281) @tsalo [DOC] Expand IBMA example (#280) @tsalo [ENH] Use PyMARE for image-based meta-analyses (#273) @tsalo [FIX] Replace NaNs in Datasets with Nones (#276) @tsalo [ENH] Support initialized and uninitialized kernels for CBMA (#275) @tsalo [ENH] Add functions to convert image types (#272) @tsalo [REF] Convert Dataset attributes to properties (#270) @tsalo [REF] Drop unimplemented annotators (#269) @tsalo [REF] Drop unimplemented parcellate module and meta-ICA workflow (#264) @tsalo [ENH] Use nearest-neighbor interpolation for masks (#258) @tsalo
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 | 0.014 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.006 |
| Open science | 0.006 | 0.006 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.427 | 0.494 |
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".