Releases: sdv-dev/SDV
Releases · sdv-dev/SDV
v1.24.1 - 2025-07-14
Bugs Fixed
- Unable to add overlapping single-table constraints to a multi-table schema - Issue #2601 by @frances-h
- Resolve DeprecationWarning (pd.api.types.is_categorical_dtype is deprecated) - Issue #2505 by @gsheni
Internal
Maintenance
v1.24.0 - 2025-06-30
New Features
- Add support for condtionally sampling column relationships and contextually anonymized columns - Issue #2582 by @frances-h
- Add support for conditional sampling null values - Issue #2581 by @frances-h
Bugs Fixed
- Unable to sample synthetic data when using timezone extraction - Issue #2589 by @pvk-developer
Internal
- Ignore Timezone Information in Datetime Constraints (Short-Term Fix) - Issue #2595 by @pvk-developer
- Add workflow to check that issues tied to PRs have a milestone - Issue #2585 by @amontanez24
v1.23.0 - 2025-06-16
New Features
- Update the wording used to get the version of SDV Community - Issue #2568 by @rwedge
- If I don't have
torch
installed, I should still be able to use SDV features that don't require torch - Issue #2551 by @rwedge
Bugs Fixed
- I should not be able to load Metadata if it contains unrecognized keys - Issue #2548 by @R-Palazzo
Internal
- Check pyproject for pre-release dependencies - Issue #2564 by @rwedge
DataProcessor
should assign'id'
sdtype to ID columns instead of'text'
- Issue #2424 by @R-Palazzo
Maintenance
- Update python set up step in workflows to use latest python version - Issue #2281 by @frances-h
v1.22.1 - 2025-06-03
Bugs Fixed
- Original metadata object passed to ProgrammableConstraint - Issue #2565 by @frances-h
v1.22.0 - 2025-06-02
New Features
- Add violin option to get_column_plot function docstrings - Issue #2554 by @amontanez24
- Allow
fit
to be an optional method forProgrammableConstraint
andSingleTableProgrammableConstraint
- Issue #2525 by @pvk-developer - Add
ProgrammableConstraint
andProgrammableSingleTableConstraint
- Issue #2513 by @frances-h - Enable single-table constraint reject sampling with multi-table synthesizers - Issue #2512 by @R-Palazzo
- Consolidate names of CAG/data patterns to
constraints
- Issue #2492 by @R-Palazzo - Add 'version' parameter to SingleTableSynthesizer.get_metadata - Issue #2484 by @pvk-developer
- Add synthesizer.validate_cag method - Issue #2471 by @gsheni
- Add CAG validation to synthesizer.validate - Issue #2470 by @R-Palazzo
- Deprecate
ScalarInequality
andScalarRange
constraints - Issue #2433 - Add CAG support to single table synthesizers - Issue #2389 by @fealho
- Add
OneHotEncoding
CAG pattern - Issue #2387 by @fealho - Add the
FixedIncrements
CAG pattern - Issue #2386 by @gsheni - Add
Range
CAG pattern - Issue #2385 by @fealho - Add
Inequality
CAG pattern - Issue #2384 by @fealho - Add
FixedCombinations
CAG pattern + add CAG base class to public - Issue #2383 by @frances-h
Bugs Fixed
- Using old style constraints should raise a
FutureWarning
- Issue #2561 by @frances-h get_constraints
for multi-table does not return single-table constraints - Issue #2559 by @R-Palazzo- Formatted columns dropped by CAG constraints may invalidate constraint - Issue #2550 by @R-Palazzo
- SDV cannot be used on a readonly filesystem - Issue #2543 by @pvk-developer
- Incorrect formatting when applying
Inequality
constraint - Issue #2524 ValueError
if conditionally sampling on a column dropped by constraints - Issue #2519 by @frances-h- Constraint hits IntCastingNanError when reverse transforming int column with nan values - Issue #2514 by @frances-h
- Inequality CAG does not respect datetime format - Issue #2495
auto_assign_transformers
errors after adding CAG pattern - Issue #2490 by @R-Palazzo- Evaluate and improve CAG pattern testing coverage - Issue #2489 by @fealho
- Inequality CAG errors out if data contains NaN values - Issue #2488 by @R-Palazzo
- Add multi-table CAG support - Issue #2487 by @frances-h
- PARSynthesizer is not aware of the sdtypes produced after pre-processing - Issue #2482 by @fealho
- Make single table CAGs backwards compatible - Issue #2446 by @fealho
v1.21.0 - 2025-05-16
New Features
- Add an API for copying the Metadata - Issue #2530 by @amontanez24
- Add an API for removing a table from the Metadata - Issue #2527 by @amontanez24
- Add an API for removing a column from the Metadata - Issue #2526 by @amontanez24
- Allow SDV to be used on a readonly filesystem - Issue #2517 by @pvk-developer
- Allow me to put in additional options when reading multiple CSV files from
CSVHandler
- Issue #2478 by @pvk-developer
Internal
v1.20.1 - 2025-05-01
Bugs Fixed
- Show a warning if I'm trying to refit/sample from a synthesizer but the metadata has changed - Issue #2463 by @pvk-developer
- Metadata auto-detection should not be creating a schema where a foreign key column is reused - Issue #2454 by @pvk-developer
- Metadata validation does not catch the case where a foreign key is reused - Issue #2453 by @pvk-developer
Maintenance
- Remove dtypes github action workflow - Issue #2475 by @gsheni
- Use IndexGenerator instead of IDGenerator from RDT - Issue #2432 by @amontanez24
v1.20.0 - 2025-04-14
New Features
- When auto-detecting metadata, add a parameter to control the foreign key detection algorithm - Issue #2456 by @amontanez24
- Provide a more descriptive error message when Regex is is not supported - Issue #2434 by @R-Palazzo
- Update transformer assignment for
id
columns - Issue #2416 by @frances-h - When in doubt, metadata auto-detection should mark columns as sdtype
categorical
- Issue #2413 by @lajohn4747 - Metadata auto-detection should find
id
columns that are not primary/foreign keys - Issue #2412 by @amontanez24
Bugs Fixed
- Metadata visualization doesn't indicate which columns are sequence key or sequence index - Issue #2411 by @lajohn4747
Internal
v1.19.0 - 2025-03-12
New Features
- Allow re-writes to metadata JSON files - Issue #2392 by @lajohn4747
Bugs Fixed
- GaussianCopula is not reporting the correct distribution name in the case of a fallback - Issue #2394 by @fealho
Internal
- Only Notify Slack on dtype Support Additions or Removals - Issue #2406 by @pvk-developer
Maintenance
v1.18.0 - 2025-02-14
New Features
- When detecting metadata from dataframes, allow me the option to turn on/off sdtype and relationship detection - Issue #2341 by @fealho
- Surface more detailed error info when detecting metadata from dataframes - Issue #2327 by @R-Palazzo
Bugs Fixed
- Conditional sampling error when using a datetime column as a context column with PAR Synthesizer - Issue #2187 by @pvk-developer
- PARSynthesizer is synthesizing integers for the
sequence_key
column when source data is text - Issue #1880 by @fealho
Maintenance
- Update our upload-artifact github action version - Issue #2370 by @amontanez24