CovsirPhy v2.20.1

Data loading and EDA

• [Fix] jhu_data.map(variable="Infected") does not show values of many countries #779

Acknowledgement

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/34

CovsirPhy v2.20.0

Data loading and EDA

• [New] Simplify legend in colored maps #747
• [New] JHUData includes population values #759
• [New] apply last date to PCRData.positive_rate() #745
• [Fix] missing dates in global Vaccinations data & colored map plot #728
• [Fix] [Performance] DataLoader.jhu() does I/O processing twice when we retrieve CSV file from server #760

ODE models and parameter estimation

• [New] alternative approach of estimation: estimate tau and ODE parameters separately #718
• [New/Revise] Show dates in accuracy graph #737
• [Fix] incorrect preset initial values of SEWIR-F model - total is not equal to population #721
• [Question/Docs] Constant population size dimensionless models #762
• [Question] Theory behind ODE system solution, Optuna optimization and estimation #698

User interface for scenario analysis

• [New] backup/restore Scenario object #741
• [Fix] Scenario.retrospective() causes IndexError #775

Forecasting

• [New] estimate delay period in prediction with grid search #731
• [Fix] Days argument in forecast - doesn't correspond to the number of prediction days #715
• [Question] Future simulation days in forecast #714
• [New/Question] Forecast plots focus #725

Acknowledgement

@Inglezos, @rebeccadavidsson and @hannanabdul55, special thanks for your discussions and pull requests!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/33?closed=1

CovsirPhy v2.19.1

Data loading

• [Fix] [outdated codes in DataLoader] COVID-19 Data Hub stops updating pre-processed data files #716

Acknowledgement

@eguidotti (author of "COVID-19 Data Hub") and @Inglezos, special thanks for your discussions!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/32?closed=1

CovsirPhy v2.19.0

S-R trend anlysis

• [Fix] too long delay period of Hungary with Scenario.estimate_delay() #693

ODE models

• [New] ModelValidator accept step_n argument to control the step number of simulation #695
• [New] value range reduction of kappa to improve parameter estimation with SIR-F #711

User interface for scenario analysis

• [New] additional metrics for evaluation #702

Forecasting

• [Fix] [Forecast] Predicted cases do not increase monotonically - Forecasted Rt negative #668

Acknowledgement

@Inglezos, special thanks for your discussions and pull requests!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/31?closed=1

CovsirPhy v2.18.0

Config/utility

• [New] Drop Python 3.6 support #647
• [New] filer to produce filenames more easily #659
• [New] covsirphy.__citation__ to show citation of CovsirPhy #674

Data loading and EDA

• [New] PCRData.subset() include daily number of tests #642
• [Fix] Scenario.register() for France data raised Segmentation fault #681

Visualization

• [New] class/function for bar plot #653
• [New] show legend at lower center in line plot #649
• [New] show legend at lower center in Scenario.trend() #651
• [New/Revise] simpify title of Scenario.trend() figure #687
• [Fix] missing area by name discrepancy in global/country-specfic map #671

S-R trend anlysis

• [New] TrendDetector class as the interface for trend analysis #662
• [Fix] [S-R trend analysis] Italy shows weird spike in estimated cases #670
• [Fix] Scenario.estimate_delay(): delay period is calculated as mean value and not representative #686

User interface for scenario analysis

• [New] Sceanrio.records(), .record_diff() and .simulate() accept variables="CFR" and "CIFR" to specify variables more easily #691
• [New] Scenario.adjust_end() to adjust the last end dates of the scenarios #666
• [Fix] past/future is not based on Scenario.today #676

Forecasting

• [New] specify end dates of future phases with Scenario.predict() #667
• [New] select variables which are used in regression #652

Acknowledgement

@Inglezos and @andrybicio, special thanks for your discussions!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/30?closed=1

CovsirPhy v2.17.0

Data loading and EDA

• [New] Global map for the number of tests/vaccinations, population and OxCGRT indicators #593
• [New] country-specific map for the number of cases/tests and population at province level #594
• [New] add "the number of vaccinated" to VaccineData as well as "the number of vaccinations" #609
• [New] method to select all country level data and all province level data in data cleaning classes #610
• [New] (JapanData) the number of vaccinated persons in Japan #612
• [Fix] subset not found in PCRData (the number of tests) #606
• [Fix] RuntimeWarning when calculating recovery period #607
• [Fix] HTTPError when downloading vaccination data and PCR test data #614
• [Fix] memory leak with .map() methods of data cleaning classes #618
• [Fix] FutureWarning in JHUdata.ensure_country_name() #635

ODE models

• [New/Revise] Deprecation of SIR-FV model because vaccination effect should be discussed with SIR-F parameters #573
• [New/Revise] deprecation of SEWIR-F parameter estimation because Rt explodes, despite RMSLE score is low #569
• [Docs] SIRS, SIRFS model not implemented #379

User interface for scenario analysis

• [New] improve interface of Scenario class to register datasets #613
• [New] property Scenario.today as the reference date to determine past/future phases #622
• [New] Scenario.records() and Scenario.records_diff() include variables of extra datasets #626
• [Fix] Scenario.history("Infected") does not include future phases #636

Forecasting

• [New] min-max scaling to improve Elastic Net regression model in parameter prediction of future phases #611
• [New] Scenario.estimate_delay() calculate the delay of any registered variables #631
• [Fix] negative Rt in parameters forecast with fit_predict() for larger delay #507

Acknowledgement

@Inglezos and @andrybicio, special thanks for your discussions!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/29?closed=1

CovsirPhy v2.16.2

Data loading

• [Fix] countries/provinces with missing values are not shown in colered maps #598
• [Fix] directory to save geometry information cannot be controlled #599
• [Fix] deprecated covsirphy.jpn_map() raises TypeError #601
• [Fix] stdout when downloading JHU dataset is not simple #603

Acknowledgement

Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/28?closed=1

CovsirPhy v2.16.1

User interface

[Fix] python kernel stops working when instantiating a scenario after upgrading to 2.16.0 version with Ananconda&pip #596

Acknowledgement

@andrybicio, special thanks for your discussions!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/27?closed=1

CovsirPhy v2.16.0

User interface

• [New] set internal methods of Term class as private methods

Data loading

• [New] [Visualization] Add colored map methods for Global, USA (states), China, Japan (prefectures) per variable #535
• [New] CountryData register the total number of provinces as country level data #555
• [Fix] data cleaning with CountryData raises ValueError with 0.0 when converting values to integers #557
• [Fix] Sourth Korea: SubsetNotFoundError raised when subseting population/OxCGRT data #552
• [Fix] ExampleData class is a child class of JHUData but raises AttributeError with .recovery_period #549
• [Fix] cleaned dataset of JHUData does not include province-level number of cases in China #588

S-R trend analysis

• [Fix] Trend analysis with Germany data fails with RuntimeError #565

Parameter estimation

• [New] change pruner of parameter estimation for speed-up #578
• [New] early stopping of parameter estimation iteration when RMSLE score not change #581
• [Fix] title overlaps in accuracy figure of parameter estimation #575
• [Fix] [Visualization] Trend plot and estimate_accuracy plot have xrange() first term set to 0 #570
• [Fix] phase numbers of stdout are not correct in parameter estimation #583

Forecasting

• [New] calculate delay value per country in parameters forecast #513

Acknowledgement

@Inglezos, @rebeccadavidsson, @andrybicio, special thanks for your discussions and pull requests!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/26?closed=1

CovsirPhy v2.15.0

Installation

• [New] Drop Python 3.6.0 - 3.6.8 support but continue Python 3.6.9 support #480
• [New] add Python 3.9 support #481
• [Fix] Installation error in Kaggle Notebook #516

Utility

• [New] Covsirphy.line_plot() accept title is None to hide title #526

Data loading

• [Fix] jhu_data.total() last record issue #483
• [Fix] PCR positive_rate() behavior for USA #490
• [Fix] PCRData.positive_rate() fails line plot when color is not specified #521
• [Fix] recovered cases in China - _subset_by_area() records selection #484

Data complement

• [New] method of Scenario to know what kind of complement was done for subset #478
• [New] method of JHUData to know what kind of complement was done for subset for each country #401
• [Fix] Partial complement recovery data of Greece shows negative infected values #488
• [Fix] Scenario.records_diff() plots weird results for last values because of inadequate complement #450
• [Fix] Estimator for Spain, France throws runtime error #500
• [Fix] un-expected full complement of JHU data (e.g. China) #514

Scenario analysis

• [New] interactive (display figures) / script mode (save figures) to Scenario class #506
• [New] Covsirphy.line_plot() accept title is None to hide title #526
• [New] Scenario.history() support for actual Confirmed as well #502
• [New] specify columns and colors in Scenario.simulate() plot #495
• [Fix] Scenario.disable() and .enable() do not accept None (i.e. all phases) #510

Acknowledgement

@Inglezos and @rebeccadavidsson, special thanks for your discussions and pull requests!
Related issues and pull requests: https://github.com/lisphilar/covid19-sir/milestone/25?closed=1