lightweight-mcnnm - Release v.1.0.2

Rewrite core estimation logic for better covariate handling and improved numerical stability.

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - Release v1.0.0

Implements additional estimation option to let user choose which fixed effects (if any) to include in estimation. Also now includes Detailed comparison to fect and causaltensor.

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - Release v0.1.5

Description of the Change

Rewrote entire code base to support jax's just-in-time-compilation (JIT), which renders use of jax.config.update("jaxdisablejit", True) obsolete. Also improves CI pipeline to include all 3 major platforms

Benefits

Significant Speedup of estimate() in most cases

Possible Drawbacks

JIT triggers recompilation of entire computation graph if the input shapes (i.e. data, in this case Y , W, and covariates) change shapes, which can be time-consuming. In cases where estimate / completematrix are called frequently on matrices of changing shapes, disabling JIT by using jax.config.update("jaxdisable_jit", True) near top of script / notebook may be useful

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - Release v0.1.4

Implements code linting and coverage testing via github actions, pre-commit hooks, and code style improvements

Full Changelog: https://github.com/tobias-schnabel/mcnnm/compare/v0.1.2...v0.1.4

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - Release v0.1.2

1. Implements performance improvements in validation methods
2. Improves documentation Full Changelog: https://github.com/tobias-schnabel/mcnnm/compare/v0.1.1...v0.1.2

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - v0.1.1

Full Changelog: https://github.com/tobias-schnabel/mcnnm/compare/v0.1.0...v0.1.1

- Python
Published by tobias-schnabel over 1 year ago

lightweight-mcnnm - v0.1.0 - Initial Release of lightweight-mcnnm

Key Features: - Efficient implementation of the MC-NNM estimator as described in Athey et al. (2021) - Utilizes JAX for improved performance and GPU compatibility - Supports various treatment assignment mechanisms - Includes unit-specific, time-specific, and unit-time specific covariates - Offers flexible validation methods for parameter selection

Highlights of this release: - Core MC-NNM estimation functionality - Support for cross-validation and holdout validation methods - Synthetic data generation for testing and examples - Comprehensive documentation and usage examples

Installation: You can install lightweight-mcnnm via pip: pip install lightweight-mcnnm

Documentation: Full documentation is available at https://mcnnm.readthedocs.io/

I welcome feedback, bug reports, and contributions from the community. Please feel free to open issues or submit pull requests on this GitHub repository.

Reference: Athey, S., Bayati, M., Doudchenko, N., Imbens, G., & Khosravi, K. (2021). Matrix Completion Methods for Causal Panel Data Models. Journal of the American Statistical Association, 116(536), 1716-1730.

- Python
Published by tobias-schnabel over 1 year ago