ModernDiD 0.2.0 Release Notes#

ModernDiD 0.2.0 introduces new features including publication-quality table generation, result extraction for estimators, GPU acceleration for continuous DiD, heterogeneous treatment effects with Bell-McCaffrey inference, and important bug fixes.

Python versions 3.11-3.13 are supported in this release.

New Features#

Publication Tables via maketables#

  • Add maketables plug-in interface to all estimator result objects for generating publication-quality LaTeX and text tables directly from results.

    (#225)

Result Extraction#

  • Expose result extraction for all estimators, providing programmatic access to estimation outputs.

    (#228)

GPU Acceleration for cont_did()#

  • Add CuPy GPU acceleration for continuous treatment DiD estimators, including device array handling, B-spline GPU tests, NPIV validation tests, and benchmarks.

    (#216)

Heterogeneous Treatment Effects#

  • Add new parameters for heterogeneous treatment effects and Bell-McCaffrey small-sample inference for did_multiplegt().

    (#222)

Enhancements#

  • Align Wald pre-test, confidence bands, and bootstrap quantiles for consistent inference across estimators.

    (#241)

  • Add error handling for NA values in input data.

    (#240)

Bug Fixes#

  • Fix clustered standard errors not propagating through att_gt().

    (#224)

  • Fix docs build failure.

    (#230)

Refactoring#

  • Update result objects for core estimators with a unified interface.

    (#229)

  • Unify synthetic data function names with gen_ prefix and add deprecated wrappers for old names.

    (#218)

  • Reorganize backend CuPy functionality.

    (#221)

Infrastructure#

  • Update test workflows to support multiple OS environments (Ubuntu, macOS, Windows).

    (#239)

  • Update installation to not include Spark by default.

    (#237)

  • Conditionally ignore validation tests on Windows.

    (#242)

Documentation#

  • Update GPU guide, installation guide, README examples, user guide, and plotting guide.

    (#238, #236, #235, #232, #231, #227, #226)

  • Enhance new estimator guide and architecture documentation.

    (#236)

  • Enhance development documentation and CI setup.

    (#223)

Contributors#

A total of 1 person contributed to this release.

  • Jordan Deklerk

Pull Requests Merged#

A total of 27 pull requests were merged for this release.

  • #242: CI: conditionally ignore validation tests on Windows

  • #241: ENH: align Wald pre-test, confidence bands, and bootstrap quantiles

  • #240: ENH: Add error handling for NA values

  • #239: CI: update test workflows to support multiple OS environments

  • #238: DOC: update GPU guide

  • #237: MAINT: update installation to not include spark by default

  • #236: DOC: enhance new estimator guide and architecture docs

  • #235: DOC: update installation guide link

  • #234: Update pre-commit hook astral-sh/ruff-pre-commit to v0.15.5

  • #233: Update github/codeql-action action to v4.32.6

  • #232: DOC: update README example to match Callaway Sant’Anna example

  • #231: DOC: update user guide and README

  • #230: BUG: fix docs failing

  • #229: REF: update result objects for core estimators

  • #228: FEAT: expose result extraction for estimators

  • #227: DOC: reorganize plotting guide

  • #226: DOC: update README to include publication tables feature

  • #225: FEAT: add maketables plug-in interface for publication-quality tables

  • #224: BUG: fix clustered standard errors not propagating through att_gt()

  • #223: DEV: enhance development documentation and CI

  • #222: FEAT: add more treatment effects and Bell-McCaffrey inference for did_multiplegt()

  • #221: MAINT: reorganize backend CuPY functionality

  • #220: Update actions/download-artifact action to v8

  • #219: Update pre-commit hook astral-sh/ruff-pre-commit to v0.15.4

  • #218: REF: unify synthetic data function names

  • #217: REF: update installation guide and add dev directory

  • #216: FEAT: add GPU acceleration for cont_did()