diff --git a/joss.06276/10.21105.joss.06276.crossref.xml b/joss.06276/10.21105.joss.06276.crossref.xml new file mode 100644 index 0000000000..a39daae07e --- /dev/null +++ b/joss.06276/10.21105.joss.06276.crossref.xml @@ -0,0 +1,377 @@ + + + + 20240403T144144-6c26a7dc7ff4522f413abab1651b18dd75cb0d25 + 20240403144144 + + JOSS Admin + admin@theoj.org + + The Open Journal + + + + + Journal of Open Source Software + JOSS + 2475-9066 + + 10.21105/joss + https://joss.theoj.org + + + + + 04 + 2024 + + + 9 + + 96 + + + + pypfilt: a particle filter for Python + + + + Robert + Moss + https://orcid.org/0000-0002-4568-2012 + + + + 04 + 03 + 2024 + + + 6276 + + + 10.21105/joss.06276 + + + http://creativecommons.org/licenses/by/4.0/ + http://creativecommons.org/licenses/by/4.0/ + http://creativecommons.org/licenses/by/4.0/ + + + + Software archive + 10.26188/25529974 + + + GitHub review issue + https://github.com/openjournals/joss-reviews/issues/6276 + + + + 10.21105/joss.06276 + https://joss.theoj.org/papers/10.21105/joss.06276 + + + https://joss.theoj.org/papers/10.21105/joss.06276.pdf + + + + + + Forecasting influenza outbreak dynamics in +Melbourne from Internet search query surveillance data + Moss + Influenza and Other Respiratory +Viruses + 4 + 10 + 10.1111/irv.12376 + 2016 + Moss, R., Zarebski, A., Dawson, P., +& McCaw, J. M. (2016). Forecasting influenza outbreak dynamics in +Melbourne from Internet search query surveillance data. Influenza and +Other Respiratory Viruses, 10(4), 314–323. +https://doi.org/10.1111/irv.12376 + + + Retrospective forecasting of the 2010–14 +Melbourne influenza seasons using multiple surveillance +systems + Moss + Epidemiology and Infection + 1 + 145 + 10.1017/S0950268816002053 + 2017 + Moss, R., Zarebski, A., Dawson, P., +& McCaw, J. M. (2017). Retrospective forecasting of the 2010–14 +Melbourne influenza seasons using multiple surveillance systems. +Epidemiology and Infection, 145(1), 156–169. +https://doi.org/10.1017/S0950268816002053 + + + Model selection for seasonal influenza +forecasting + Zarebski + Infectious Disease Modelling + 1 + 2 + 10.1016/j.idm.2016.12.004 + 2468-0427 + 2017 + Zarebski, A. E., Dawson, P., McCaw, +J. M., & Moss, R. (2017). Model selection for seasonal influenza +forecasting. Infectious Disease Modelling, 2(1), 56–70. +https://doi.org/10.1016/j.idm.2016.12.004 + + + Epidemic forecasts as a tool for public +health: Interpretation and (re)calibration + Moss + Australian and New Zealand Journal of Public +Health + 1 + 42 + 10.1111/1753-6405.12750 + 2018 + Moss, R., Fielding, J. E., Franklin, +L. J., Stephens, N., McVernon, J., Dawson, P., & McCaw., J. M. +(2018). Epidemic forecasts as a tool for public health: Interpretation +and (re)calibration. Australian and New Zealand Journal of Public +Health, 42(1), 69–76. +https://doi.org/10.1111/1753-6405.12750 + + + Disease forecasting system takes out National +Innovation Awards + Pyne + Media release + 2018 + Pyne, C. (2018). Disease forecasting +system takes out National Innovation Awards. In Media release. +Australian Department of Defence. +https://www.minister.defence.gov.au/media-releases/2018-05-11/disease-forecasting-system-takes-out-national-innovation-awards + + + Accounting for healthcare-seeking behaviours +and testing practices in real-time influenza forecasts + Moss + Tropical Medicine and Infectious +Disease + 1 + 4 + 10.3390/tropicalmed4010012 + 2019 + Moss, R., Zarebski, A. E., Carlson, +S. J., & McCaw., J. M. (2019). Accounting for healthcare-seeking +behaviours and testing practices in real-time influenza forecasts. +Tropical Medicine and Infectious Disease, 4(1), 12. +https://doi.org/10.3390/tropicalmed4010012 + + + Anatomy of a seasonal influenza epidemic +forecast + Moss + Communicable Diseases +Intelligence + 43 + 10.33321/cdi.2019.43.7 + 2019 + Moss, R., Zarebski, A. E., Dawson, +P., Franklin, L. J., Birrell, F. A., & McCaw., J. M. (2019). Anatomy +of a seasonal influenza epidemic forecast. Communicable Diseases +Intelligence, 43, 1–14. +https://doi.org/10.33321/cdi.2019.43.7 + + + Early analysis of the Australian COVID-19 +epidemic + Price + eLife + 9 + 10.7554/eLife.58785 + 2020 + Price, D. J., Shearer, F. M., Meehan, +M. T., McBryde, E. S., Moss, R., Golding, N., Conway, E. J., Dawson, P., +Cromer, D., Wood, J., Abbott, S., McVernon, J., & McCaw, J. M. +(2020). Early analysis of the Australian COVID-19 epidemic. eLife, 9, +e58785. https://doi.org/10.7554/eLife.58785 + + + Infectious disease pandemic planning and +response: Incorporating decision analysis + Shearer + PLOS Medicine + 17 + 10.1371/journal.pmed.1003018 + 2020 + Shearer, F. M., Moss, R., McVernon, +J., Ross, J. V., & McCaw, J. M. (2020). Infectious disease pandemic +planning and response: Incorporating decision analysis. PLOS Medicine, +17, e1003018. +https://doi.org/10.1371/journal.pmed.1003018 + + + Coronavirus disease model to inform +transmission-reducing measures and health system preparedness, +Australia + Moss + Emerging Infectious Diseases + 12 + 26 + 10.3201/eid2612.202530 + 2020 + Moss, R., Wood, J., Brown, D., +Shearer, F. M., Black, A. J., Glass, K., Cheng, A. C., McCaw, J. M., +& McVernon, J. (2020). Coronavirus disease model to inform +transmission-reducing measures and health system preparedness, +Australia. Emerging Infectious Diseases, 26(12), 2844–2853. +https://doi.org/10.3201/eid2612.202530 + + + Forecasting COVID-19 activity in Australia to +support pandemic response: May to October 2020 + Moss + Scientific Reports + 1 + 13 + 10.1038/s41598-023-35668-6 + 2023 + Moss, R., Price, D. J., Golding, N., +Dawson, P., McVernon, J., Hyndman, R. J., Shearer, F. M., & McCaw, +J. M. (2023). Forecasting COVID-19 activity in Australia to support +pandemic response: May to October 2020. Scientific Reports, 13(1). +https://doi.org/10.1038/s41598-023-35668-6 + + + Monte Carlo filter and smoother for +non-Gaussian nonlinear state space models + Kitagawa + Journal of Computational and Graphical +Statistics + 1 + 5 + 10.1080/10618600.1996.10474692 + 1996 + Kitagawa, G. (1996). Monte Carlo +filter and smoother for non-Gaussian nonlinear state space models. +Journal of Computational and Graphical Statistics, 5(1), 1–25. +https://doi.org/10.1080/10618600.1996.10474692 + + + Sequential Monte Carlo methods in +practice + 10.1007/978-1-4757-3437-9 + 978-1-4419-2887-0 + 2001 + Doucet, A., Freitas, N. de, & +Gordon, N. (Eds.). (2001). Sequential Monte Carlo methods in practice +(1st ed.). Springer. +https://doi.org/10.1007/978-1-4757-3437-9 + + + Improving regularised particle +filters + Musso + Sequential Monte Carlo methods in +practice + 10.1007/978-1-4757-3437-9_12 + 2001 + Musso, C., Oudjane, N., & Le +Gland, F. (2001). Improving regularised particle filters. In Sequential +Monte Carlo methods in practice (pp. 247–271). Springer. +https://doi.org/10.1007/978-1-4757-3437-9_12 + + + An introduction to Sequential Monte +Carlo + Chopin + 10.1007/978-3-030-47845-2 + 2020 + Chopin, N., & Papaspiliopoulos, +O. (2020). An introduction to Sequential Monte Carlo. Springer. +https://doi.org/10.1007/978-3-030-47845-2 + + + Forecasting: Principles and +practice + Hyndman + 2021 + Hyndman, R. J., & Athanasopoulos, +G. (2021). Forecasting: Principles and practice (3rd ed.). OTexts. +https://otexts.com/fpp3 + + + PyMC: A modern, and comprehensive +probabilistic programming framework in Python + Abril-Pla + PeerJ Computer Science + 9 + 10.7717/peerj-cs.1516 + 2376-5992 + 2023 + Abril-Pla, O., Andreani, V., Carroll, +C., Dong, L., Fonnesbeck, C. J., Kochurov, M., Kumar, R., Lao, J., +Luhmann, C. C., Martin, O. A., Osthege, M., Vieira, R., Wiecki, T., +& Zinkov, R. (2023). PyMC: A modern, and comprehensive probabilistic +programming framework in Python. PeerJ Computer Science, 9, e1516. +https://doi.org/10.7717/peerj-cs.1516 + + + Pyro: Deep universal probabilistic +programming + Bingham + Journal of Machine Learning +Research + 20 + 2019 + Bingham, E., Chen, J. P., Jankowiak, +M., Obermeyer, F., Pradhan, N., Karaletsos, T., Singh, R., Szerlip, P. +A., Horsfall, P., & Goodman, N. D. (2019). Pyro: Deep universal +probabilistic programming. Journal of Machine Learning Research, 20, +28:1–28:6. +http://jmlr.org/papers/v20/18-403.html + + + Dstl/stone-soup: v1.1 release + Hiscocks + 10.5281/zenodo.8308177 + 2023 + Hiscocks, S., Harrald, O., Barr, J., +Perree, N., Vladimirov, L., Green, R., Rosoman, O., etfrogers-dstl, +idorrington-dstl, Glover, T., Hunter, E., Wright, J., gawebb-dstl, +Harris, M., Fraser, B., spike, Pritchett, H., jjosborne-dstl, Carniglia, +P., … Hiles, J. (2023). Dstl/stone-soup: v1.1 release (Version v1.1). +Zenodo. https://doi.org/10.5281/zenodo.8308177 + + + FilterPy - Kalman filters and other optimal +and non-optimal estimation filters in Python + Labbe + GitHub repository + 2022 + Labbe, R. R. (2022). FilterPy - +Kalman filters and other optimal and non-optimal estimation filters in +Python. In GitHub repository. GitHub. +https://github.com/rlabbe/filterpy + + + SMCPy - Sequential Monte Carlo with +Python + Leser + GitHub repository + 2023 + Leser, P., & Wang, M. (2023). +SMCPy - Sequential Monte Carlo with Python. In GitHub repository. +GitHub. https://github.com/nasa/SMCPy + + + + + + diff --git a/joss.06276/10.21105.joss.06276.jats b/joss.06276/10.21105.joss.06276.jats new file mode 100644 index 0000000000..03b75fafe6 --- /dev/null +++ b/joss.06276/10.21105.joss.06276.jats @@ -0,0 +1,668 @@ + + +
+ + + + +Journal of Open Source Software +JOSS + +2475-9066 + +Open Journals + + + +6276 +10.21105/joss.06276 + +pypfilt: a particle filter for Python + + + +https://orcid.org/0000-0002-4568-2012 + +Moss +Robert + + + + + +Melbourne School of Population and Global Health, The +University of Melbourne, Australia + + + + +20 +12 +2023 + +9 +96 +6276 + +Authors of papers retain copyright and release the +work under a Creative Commons Attribution 4.0 International License (CC +BY 4.0) +2022 +The article authors + +Authors of papers retain copyright and release the work under +a Creative Commons Attribution 4.0 International License (CC BY +4.0) + + + +Python +particle filter +forecasting + + + + + + Summary +

Mathematical models are used to simulate real-world systems in many + scientific fields. These models can be fitted to real-time data, and + used to generate probabilistic forecasts that + describe how the system will behave in the future and convey the + uncertainty in these predictions. Particle filters + are a class of Sequential Monte Carlo (SMC) methods + (Doucet + et al., 2001) that have been used in many scientific fields for + real-time forecasting, with the COVID-19 pandemic being one of the + most recent and high-profile examples. These methods can be used to + estimate model parameters and state as new data become available + (“online estimation”).

+

pypfilt is a Python package for online + estimation and forecasting that implements several particle filters. + It was developed to enable real-time seasonal influenza forecasting in + Australia, for which we won two national innovation awards + (Pyne, + 2018), and played a key role in generating forecasts that have + supported Australia’s COVID-19 response + (Moss + et al., 2023). The package is deliberately generic and readily + applicable to other domains.

+ + + Statement of need +

Particle filters are provided by Python packages that implement a + range of inference methods, such as PyMC + (Abril-Pla + et al., 2023), pyro + (Bingham + et al., 2019), and stonesoup + (Hiscocks + et al., 2023); accompany textbooks, such as + particles + (Chopin + & Papaspiliopoulos, 2020) and + filterpy + (Labbe, + 2022); and focus on other applications, such as + SMCPy + (Leser + & Wang, 2023). However, pypfilt + appears to be unique in supporting all of the following: (a) online + state estimation for time-series forecasting; (b) arbitrary + state-space models; (c) non-analytic likelihood functions; (d) control + over memory usage for large-scale applications; (e) a declarative + approach for defining and running forecasts; and (f) reproducible + results. In brief:

+ + +

Forecasts are generated efficiently. Repeated + calculations are avoided when new data are received (online + estimation), even when previous data are updated or corrected, due + to a sophisticated caching system.

+ + +

Complex models are easily defined. Models only + need to define the state vector, with arbitrarily nested fields + and dimensions, and define an update rule for the state + vector.

+ + +

Likelihood functions are unconstrained. They + do not need to be differentiable, and can inspect both the current + state and any previous states.

+ + +

Memory usage is flexible. The particle filter + can retain only a sliding window of previous states, and only + record states at coarse intervals, so that forecasts with large + numbers of particles and/or very small time-steps do not exceed + available memory.

+ + +

Forecasts are simple to define and run. + Forecasts are defined in plain-text configuration files, enforcing + a clear separation between model implementation and experiment + (e.g., choice of prior distributions, input data, particle filter + settings).

+ + +

Reproducibility is ensured. Output files + include all data and metadata required to reproduce the original + results.

+ + +

Additional features include supporting scalar and calendar time + scales, providing a range of resampling strategies + (Kitagawa, + 1996) and summary statistics, post-regularisation + (Musso + et al., 2001), and measuring forecast performance with + Continuous Ranked Probability Scores (CRPS) + (Hyndman + & Athanasopoulos, 2021). It also supports scenario + modelling — simulating from multiple prior distributions and + comparing matching particles in each ensemble — which provides + additional decision-support capabilities beyond those provided by + real-time forecasts + (Moss + et al., 2020; + Shearer + et al., 2020).

+

A suite of almost 150 + test + cases (comprising more than 6,000 lines of code) runs + automatically every time the code is updated. This includes tests + which verify that outputs are reproducible, and tests which verify + that outputs are identical whether or not the particle filter resumed + from a previously-cached state.

+ + + Availability and usage +

This is available as a Python package on + PyPI + and can be installed using pip. It deliberately + supports older versions of Python ( + + ≥3.8), + NumPy ( + + ≥1.17), + and SciPy ( + + ≥1.4). + The documentation is available at + https://pypfilt.readthedocs.io/, + and includes a + Getting + Started tutorial that demonstrates how to construct a + model, fit it to data, and generate forecasts (see + [fig:example]). + In this tutorial we use the Lorenz-63 system of ordinary differential + equations (which has chaotic solutions) to show how + post-regularisation can greatly improve forecast performance (see + [fig:perf]), and to + highlight how easy it is to define and run a suite of forecasts.

+ +

An example forecast at time + + + t=20 + for the Lorenz-63 system. This figure was generated using the + pypfilt.plot + module.

+ + + + + Ongoing research projects +

Beginning in 2015, pypfilt was developed to + support real-time seasonal influenza forecasting in Australia + (Moss + et al., 2016, + 2017, + 2018; + Moss, + Zarebski, Carlson, et al., 2019; + Moss, + Zarebski, Dawson, et al., 2019; + Zarebski + et al., 2017), and has been used to support the Australian + Government’s response to COVID-19 + (Moss + et al., 2023; + Price + et al., 2020).

+ +

An example of using Continuous Ranked Probability Scores + (CPRS) to compare forecast performance. Post-regularisation improves + performance by 76.7% in this + example.

+ + + + + Acknowledgements +

Peter Dawson, James M. McCaw, David J. Price, and Alexander E. + Zarebski contributed helpful comments and suggestions. Package + development was supported by the Defence Science and Technology Group + project “Bioterrorism Preparedness Strategic Research Initiative + 07/301”, and by Australian National Health and Medical Research + Council (NHMRC) Centres for Research Excellence (PRISM, 1078068; + APPRISE, 1116530; SPECTRUM, 1170960). RM was supported by an NHMRC + APPRISE Research Fellowship (1116530).

+ + + + + + + + MossRobert + ZarebskiAlex + DawsonPeter + McCawJames M. + + Forecasting influenza outbreak dynamics in Melbourne from Internet search query surveillance data + Influenza and Other Respiratory Viruses + 2016 + 10 + 4 + 10.1111/irv.12376 + 314 + 323 + + + + + + MossRobert + ZarebskiAlex + DawsonPeter + McCawJames M. + + Retrospective forecasting of the 2010–14 Melbourne influenza seasons using multiple surveillance systems + Epidemiology and Infection + 2017 + 145 + 1 + 10.1017/S0950268816002053 + 156 + 169 + + + + + + ZarebskiAlexander E. + DawsonPeter + McCawJames M. + MossRobert + + Model selection for seasonal influenza forecasting + Infectious Disease Modelling + Elsevier BV + 201702 + 2 + 1 + 2468-0427 + 10.1016/j.idm.2016.12.004 + 56 + 70 + + + + + + MossRobert + FieldingJames E. + FranklinLucinda J. + StephensNicola + McVernonJodie + DawsonPeter + McCaw.James M. + + Epidemic forecasts as a tool for public health: Interpretation and (re)calibration + Australian and New Zealand Journal of Public Health + 2018 + 42 + 1 + 10.1111/1753-6405.12750 + 69 + 76 + + + + + + PyneChristopher + + Disease forecasting system takes out National Innovation Awards + Media release + Australian Department of Defence + 2018 + https://www.minister.defence.gov.au/media-releases/2018-05-11/disease-forecasting-system-takes-out-national-innovation-awards + + + + + + MossRobert + ZarebskiAlexander E. + CarlsonSandra J. + McCaw.James M. + + Accounting for healthcare-seeking behaviours and testing practices in real-time influenza forecasts + Tropical Medicine and Infectious Disease + 2019 + 4 + 1 + 10.3390/tropicalmed4010012 + 12 + + + + + + + MossRobert + ZarebskiAlexander E. + DawsonPeter + FranklinLucinda J. + BirrellFrances A. + McCaw.James M. + + Anatomy of a seasonal influenza epidemic forecast + Communicable Diseases Intelligence + 2019 + 43 + 10.33321/cdi.2019.43.7 + 1 + 14 + + + + + + PriceDavid J. + ShearerFreya M. + MeehanMichael T. + McBrydeEmma S. + MossRobert + GoldingNick + ConwayEamon J. + DawsonPeter + CromerDeborah + WoodJames + AbbottSam + McVernonJodie + McCawJames M. + + Early analysis of the Australian COVID-19 epidemic + eLife + 2020 + 9 + 10.7554/eLife.58785 + e58785 + + + + + + + ShearerFreya M. + MossRobert + McVernonJodie + RossJoshua V. + McCawJames M. + + Infectious disease pandemic planning and response: Incorporating decision analysis + PLOS Medicine + 2020 + 17 + 10.1371/journal.pmed.1003018 + e1003018 + + + + + + + MossRobert + WoodJames + BrownDamien + ShearerFreya M. + BlackAndrew J. + GlassKathryn + ChengAllen C. + McCawJames M. + McVernonJodie + + Coronavirus disease model to inform transmission-reducing measures and health system preparedness, Australia + Emerging Infectious Diseases + Centers for Disease Control; Prevention (CDC) + 202012 + 26 + 12 + https://doi.org/10.3201/eid2612.202530 + 10.3201/eid2612.202530 + 2844 + 2853 + + + + + + MossRobert + PriceDavid J. + GoldingNick + DawsonPeter + McVernonJodie + HyndmanRob J. + ShearerFreya M. + McCawJames M. + + Forecasting COVID-19 activity in Australia to support pandemic response: May to October 2020 + Scientific Reports + Springer Science; Business Media LLC + 202305 + 13 + 1 + 10.1038/s41598-023-35668-6 + + + + + + KitagawaGenshiro + + Monte Carlo filter and smoother for non-Gaussian nonlinear state space models + Journal of Computational and Graphical Statistics + 1996 + 5 + 1 + 10.1080/10618600.1996.10474692 + 1 + 25 + + + + + Sequential Monte Carlo methods in practice + + DoucetArnaud + FreitasNando de + GordonNeil + + Springer + New York + 2001 + 1st + 978-1-4419-2887-0 + 10.1007/978-1-4757-3437-9 + + + + + + MussoChristian + OudjaneNadia + Le GlandFrançois + + Improving regularised particle filters + Sequential Monte Carlo methods in practice + Springer + 2001 + 10.1007/978-1-4757-3437-9_12 + 247 + 271 + + + + + + ChopinNicolas + PapaspiliopoulosOmiros + + An introduction to Sequential Monte Carlo + Springer + Basel + 2020 + 10.1007/978-3-030-47845-2 + + + + + + HyndmanRob J. + AthanasopoulosGeorge + + Forecasting: Principles and practice + OTexts + Melbourne, Australia + 2021 + 3rd + https://otexts.com/fpp3 + + + + + + Abril-PlaOriol + AndreaniVirgile + CarrollColin + DongLarry + FonnesbeckChristopher J. + KochurovMaxim + KumarRavin + LaoJunpeng + LuhmannChristian C. + MartinOsvaldo A. + OsthegeMichael + VieiraRicardo + WieckiThomas + ZinkovRobert + + PyMC: A modern, and comprehensive probabilistic programming framework in Python + PeerJ Computer Science + PeerJ + 202309 + 9 + 2376-5992 + 10.7717/peerj-cs.1516 + e1516 + + + + + + + BinghamEli + ChenJonathan P. + JankowiakMartin + ObermeyerFritz + PradhanNeeraj + KaraletsosTheofanis + SinghRohit + SzerlipPaul A. + HorsfallPaul + GoodmanNoah D. + + Pyro: Deep universal probabilistic programming + Journal of Machine Learning Research + 2019 + 20 + http://jmlr.org/papers/v20/18-403.html + 28:1 + 28:6 + + + + + + HiscocksSteven + HarraldOliver + BarrJordi + PerreeNicola + VladimirovLyudmil + GreenRichard + RosomanOliver + etfrogers-dstl + idorrington-dstl + GloverTimothy + HunterEmily + WrightJames + gawebb-dstl + HarrisMichael + FraserBenjamin + spike + PritchettHenry + jjosborne-dstl + CarnigliaPeter + cje20a + lflaherty-dstl + KirklandDavid + rg + DaviesRichard + snaylor20 + CampbellMark + dcrew-dstl + dlast-dstl + ThomasSam + HilesJohn + + Dstl/stone-soup: v1.1 release + Zenodo + 202309 + https://doi.org/10.5281/zenodo.8308177 + 10.5281/zenodo.8308177 + + + + + + LabbeRoger R. + + FilterPy - Kalman filters and other optimal and non-optimal estimation filters in Python + GitHub repository + GitHub + 2022 + https://github.com/rlabbe/filterpy + + + + + + LeserPatrick + WangMichael + + SMCPy - Sequential Monte Carlo with Python + GitHub repository + GitHub + 2023 + https://github.com/nasa/SMCPy + + + + +
diff --git a/joss.06276/10.21105.joss.06276.pdf b/joss.06276/10.21105.joss.06276.pdf new file mode 100644 index 0000000000..7db74c762f Binary files /dev/null and b/joss.06276/10.21105.joss.06276.pdf differ diff --git a/joss.06276/media/lorenz63_crps_comparison.png b/joss.06276/media/lorenz63_crps_comparison.png new file mode 100644 index 0000000000..340d6a2a52 Binary files /dev/null and b/joss.06276/media/lorenz63_crps_comparison.png differ diff --git a/joss.06276/media/lorenz63_forecast_regularised.png b/joss.06276/media/lorenz63_forecast_regularised.png new file mode 100644 index 0000000000..d0eef3d059 Binary files /dev/null and b/joss.06276/media/lorenz63_forecast_regularised.png differ