# PyTimeVar: A Python Package for Trending Time-Varying Time Series Models
<!-- badges: start -->
[](https://www.gnu.org/licenses/gpl-3.0)
<!-- badges: end -->
Authors: Mingxuan Song (m3.song@student.vu.nl, Vrije Universiteit Amsterdam), Bernhard van der Sluis (vandersluis@ese.eur.nl, Erasmus Universiteit Rotterdam), and Yicong Lin (yc.lin@vu.nl, Vrije Universiteit Amsterdam & Tinbergen Institute)
## Purpose of the package
The PyTimeVar package offers state-of-the-art estimation and statistical inference methods for time series regression models with flexible trends and/or time-varying coefficients. The package implements nonparametric estimation along with multiple recently proposed bootstrap-assisted inference methods. Pointwise confidence intervals and simultaneous bands of parameter curves via bootstrap can be easily obtained using user-friendly commands. The package also includes four commonly used methods for modeling trends and time-varying relationships: boosted Hodrick-Prescot filter, power-law trend models, state-space models, and score-driven models. This allows users to compare different approaches within a unified environment.
The package is built upon several papers and books. We list the key references below.
### Local linear kernel estimation and bootstrap inference
Friedrich and Lin (2024) (doi: https://doi.org/10.1016/j.jeconom.2022.09.004);
Lin et al. (2024) (doi: https://doi.org/10.1080/10618600.2024.2403705);
Friedrich et al. (2020) (doi: https://doi.org/10.1016/j.jeconom.2019.05.006);
Smeekes and Urbain (2014) (doi: https://doi.org/10.26481/umagsb.2014008)
Zhou and Wu (2010) (doi: https://doi.org/10.1111/j.1467-9868.2010.00743.x);
Bühlmann (1998) (doi: https://doi.org/10.1214/aos/1030563978);
### Boosted HP filter
Mei et al. (2024) (doi: doi: https://doi.org/10.1002/jae.3086);
Biswas et al. (2024) (doi: https://doi.org/10.1080/07474938.2024.2380704);
Phillips and Shi (2021) (doi: https://doi.org/10.1111/iere.12495);
### Power-law trend models
Lin and Reuvers (2024) (https://tinbergen.nl/discussion-paper/6214/22-092-iii-cointegrating-polynomial-regressions-with-power-law-trends-environmental-kuznets-curve-or-omitted-time-effects);
Robinson (2012) (doi: https://doi.org/10.3150/10-BEJ349);
### State-space models
Durbin and Koopman (2012) (doi: https://doi.org/10.1093/acprof:oso/9780199641178.001.0001)
### Score-drive models
Creal et al. (2013) (doi: https://doi.org/10.1002/jae.1279);
Harvey (2013) (doi: https://doi.org/10.1017/CBO9781139540933);
## Features
- Nonparametric estimation of time-varying time series models, along with various bootstrap-assisted methods for inference, including local blockwise wild bootstrap, wild bootstrap, sieve bootstrap, sieve wild bootstrap, autoregressive wild bootstrap
- Alternative estimation methods for modeling trend and time-varying relationships, including boosted HP filter, power-law trend models, state-space, and score-driven models.
- Unified framework for comparison of methods.
- Multiple datasets for illustration.
## Getting started
The PyTimeVar can implemented as a PyPI package. To download the package in your Python environment, use the following command:
```python
pip install PyTimeVar
```
## Support
The documentation of the package can be found at the GitHub repository https://github.com/bpvand/PyTimeVar, and ReadTheDocs https://pytimevar.readthedocs.io/en/latest/.
For any questions or feedback regarding the PyTimeVar package, please feel free to contact the authors via email:
m3.song@student.vu.nl;
vandersluis@ese.eur.nl;
yc.lin@vu.nl.
Raw data
{
"_id": null,
"home_page": "https://github.com/bpvand/PyTimeVar",
"name": "PyTimeVar",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": null,
"keywords": "time-varying, bootstrap, nonparametric estimation, filtering",
"author": "Mingxuan Song, Bernhard van der Sluis, Yicong Lin",
"author_email": "678270ms@eur.nl, vandersluis@ese.eur.nl, yc.lin@vu.nl",
"download_url": "https://files.pythonhosted.org/packages/f0/db/5e84ca0c7984b84de11e613380a4e9734d25b886de462aa103453a61706b/pytimevar-1.0.0.tar.gz",
"platform": null,
"description": "# PyTimeVar: A Python Package for Trending Time-Varying Time Series Models\r\n<!-- badges: start -->\r\n[](https://www.gnu.org/licenses/gpl-3.0)\r\n\r\n\r\n<!-- badges: end -->\r\n\r\nAuthors: Mingxuan Song (m3.song@student.vu.nl, Vrije Universiteit Amsterdam), Bernhard van der Sluis (vandersluis@ese.eur.nl, Erasmus Universiteit Rotterdam), and Yicong Lin (yc.lin@vu.nl, Vrije Universiteit Amsterdam & Tinbergen Institute)\r\n\r\n## Purpose of the package\r\n\r\nThe PyTimeVar package offers state-of-the-art estimation and statistical inference methods for time series regression models with flexible trends and/or time-varying coefficients. The package implements nonparametric estimation along with multiple recently proposed bootstrap-assisted inference methods. Pointwise confidence intervals and simultaneous bands of parameter curves via bootstrap can be easily obtained using user-friendly commands. The package also includes four commonly used methods for modeling trends and time-varying relationships: boosted Hodrick-Prescot filter, power-law trend models, state-space models, and score-driven models. This allows users to compare different approaches within a unified environment.\r\n\r\nThe package is built upon several papers and books. We list the key references below.\r\n\r\n### Local linear kernel estimation and bootstrap inference\r\nFriedrich and Lin (2024) (doi: https://doi.org/10.1016/j.jeconom.2022.09.004);\r\nLin et al. (2024) (doi: https://doi.org/10.1080/10618600.2024.2403705);\r\nFriedrich et al. (2020) (doi: https://doi.org/10.1016/j.jeconom.2019.05.006);\r\nSmeekes and Urbain (2014) (doi: https://doi.org/10.26481/umagsb.2014008)\r\nZhou and Wu (2010) (doi: https://doi.org/10.1111/j.1467-9868.2010.00743.x);\r\nB\u00c3\u00bchlmann (1998) (doi: https://doi.org/10.1214/aos/1030563978);\r\n\r\n\r\n### Boosted HP filter\r\nMei et al. (2024) (doi: doi: https://doi.org/10.1002/jae.3086);\r\nBiswas et al. (2024) (doi: https://doi.org/10.1080/07474938.2024.2380704);\r\nPhillips and Shi (2021) (doi: https://doi.org/10.1111/iere.12495);\r\n\r\n\r\n### Power-law trend models\r\nLin and Reuvers (2024) (https://tinbergen.nl/discussion-paper/6214/22-092-iii-cointegrating-polynomial-regressions-with-power-law-trends-environmental-kuznets-curve-or-omitted-time-effects);\r\nRobinson (2012) (doi: https://doi.org/10.3150/10-BEJ349);\r\n\r\n\r\n### State-space models\r\nDurbin and Koopman (2012) (doi: https://doi.org/10.1093/acprof:oso/9780199641178.001.0001)\r\n\r\n### Score-drive models\r\nCreal et al. (2013) (doi: https://doi.org/10.1002/jae.1279);\r\nHarvey (2013) (doi: https://doi.org/10.1017/CBO9781139540933);\r\n\r\n## Features\r\n\r\n- Nonparametric estimation of time-varying time series models, along with various bootstrap-assisted methods for inference, including local blockwise wild bootstrap, wild bootstrap, sieve bootstrap, sieve wild bootstrap, autoregressive wild bootstrap\r\n- Alternative estimation methods for modeling trend and time-varying relationships, including boosted HP filter, power-law trend models, state-space, and score-driven models.\r\n- Unified framework for comparison of methods.\r\n- Multiple datasets for illustration.\r\n\r\n## Getting started\r\n\r\nThe PyTimeVar can implemented as a PyPI package. To download the package in your Python environment, use the following command:\r\n```python\r\npip install PyTimeVar\r\n```\r\n\r\n## Support\r\nThe documentation of the package can be found at the GitHub repository https://github.com/bpvand/PyTimeVar, and ReadTheDocs https://pytimevar.readthedocs.io/en/latest/.\r\n\r\nFor any questions or feedback regarding the PyTimeVar package, please feel free to contact the authors via email: \r\nm3.song@student.vu.nl; \r\nvandersluis@ese.eur.nl; \r\nyc.lin@vu.nl.\r\n",
"bugtrack_url": null,
"license": "GPLv3+",
"summary": "The PyTimeVar package offers state-of-the-art estimation and statistical inference methods for time series regression models with flexible trends and/or time- varying coefficients.",
"version": "1.0.0",
"project_urls": {
"Homepage": "https://github.com/bpvand/PyTimeVar"
},
"split_keywords": [
"time-varying",
" bootstrap",
" nonparametric estimation",
" filtering"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "da721e560f0a6d256a5b14e18329e56d95e73526e15ea04bf99d227500db7af6",
"md5": "d46c6af205e858eca53bf945e9bc2a3d",
"sha256": "46631e92ccab64c0a4336e0b27a8b755a1d193205ba8b20f5fefe7bfc1a333e5"
},
"downloads": -1,
"filename": "PyTimeVar-1.0.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "d46c6af205e858eca53bf945e9bc2a3d",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 232623,
"upload_time": "2024-09-20T13:29:54",
"upload_time_iso_8601": "2024-09-20T13:29:54.663068Z",
"url": "https://files.pythonhosted.org/packages/da/72/1e560f0a6d256a5b14e18329e56d95e73526e15ea04bf99d227500db7af6/PyTimeVar-1.0.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "f0db5e84ca0c7984b84de11e613380a4e9734d25b886de462aa103453a61706b",
"md5": "a80b1098ab8c44a40f77f7fb64e7a1ef",
"sha256": "c95963cab8bfafdf2d66a845da2a024967fa16d44d2e3a6d42c4d3c2e79ab415"
},
"downloads": -1,
"filename": "pytimevar-1.0.0.tar.gz",
"has_sig": false,
"md5_digest": "a80b1098ab8c44a40f77f7fb64e7a1ef",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 229756,
"upload_time": "2024-09-20T13:29:59",
"upload_time_iso_8601": "2024-09-20T13:29:59.121063Z",
"url": "https://files.pythonhosted.org/packages/f0/db/5e84ca0c7984b84de11e613380a4e9734d25b886de462aa103453a61706b/pytimevar-1.0.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-09-20 13:29:59",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "bpvand",
"github_project": "PyTimeVar",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "pytimevar"
}
JSON
