CovsirPhy v.2.25.0

Breaking changes

• [New] accept city-level data in analysis #1041
• [New] change of data loading scheme: download datasets of selected location #1045
• [New] DataEngineer class for data cleaning, transforming and complementing #1064

Most of classes were deprecated and new classes were created. Please refer to Documentation for tutorials of the new classes.

Others and documentation

• [Fix] address format of county and province: province/country, not country/province #1059
• [Fix] too much time to retrieve dataset from COVID-19 Data Hub #986
• [Revise] stdout when downloading datasets from CovsirPhy/data project #1051
• [Data] failed in ODE parameter estimation with Japan data: IndexError #1087
• [Question/Docs] how to access S-R trend analysis statistical details #894

Acknowledgement

Special thanks for users and developers!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/44?closed=1

CovsirPhy v2.24.0

Forecasting

• [New] ODE parameter prediction with AutoML #1010
• [New/Revise] Scenario.fit() will be merged to Scenario.predict() #1006
• [Discuss] time series forecasting without selecting delay period #792

Scenario analysis

• [New] find out the best/worst scenarios #1012
• [New] rename scenario name #1022
• [New] delete scenarios with regular expressions #1023
• [Question/Docs] Paramters to tweek #999

Acknowledgement

Special thanks for users and developers!

• hfatima1

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/43?closed=1

CovsirPhy v2.23.1

Data loading

• [Fix] vaccination data (1st, 2nd) was removed from total vaccination data mistakenly #996

Acknowledgement

Special thanks for users and developers!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/42?closed=1

CovsirPhy v2.23.0

Config

• [Revise] update the last year in citation, 2020-2021 to 2020-2022 #970

Data loading

• [New] (Global data) booster shots of vaccination is not included in "Vaccinations" data #991
• [New] (Japan data) 3rd shots of vaccination is not included in "Vaccinations" data #969
• [Fix] ExampleData, TypeError: unsupported operand type(s) for -: 'numpy.ndarray' and 'Timestamp' #982
• [Fix] Database Lock: numpy.core._exceptions._ArrayMemoryError #983
• [Fix] DataLoader(update_interval=None).japan() raises TypeError #988
• [Question/Docs] How to resolve SubsetNotFoundError: "No records with 'Recovered > 0' in 10000, US were found." for cities with less pandemic prevalence? #974

Acknowledgement

Special thanks for users and developers!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/40?closed=1

CovsirPhy v2.22.2

Scenario analysis

• [Fix] fit_predict() raises TypeError when zeros are used as parameter values #954
• [Fix] class Scenario estimate_accuracy() and fit() returns KeyError #957

Acknowledgement

Special thanks for users and developers!

• @baluballa

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/41?closed=1

CovsirPhy v2.22.1

Data loading

• [Fix] MobilityData.map() when country argument not used raises TypeError #916
• [Fix] JHUData.records(): complement is done after subsetting with start date #921
• [Fix] JHUData.subset(country="China") raises SubsetNotFoundError #951

Acknowledgement

Special thanks for users and developers!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/39?closed=1

CovsirPhy v2.22.0

Data loading

• [New] Extending support for County level COVID-19 Impact Assessment #851
• [New] DataLoader.collect() to collect all datasets for scenario analysis #853
• [New] Specify strftime format with CountryData.cleaned(date_format) when we use local dataset (Fix: Using Own Dataset Not Work Anymore) #856
• [New] dataset regarding mobility data #862
• [New] flexible configuration of variables in OxCGRTData and MobilityData #899
• [New] JHUData.cleaned() includes ISO3 codes, the number of infected cases and population #900
• [New] OxCGRTData accepts province level data #901
• [Fix] exclude international aggregates and partial subnational data from automatically loaded vaccine data #891
• [Fix] DeprecationWarning with dependencies and ColoredMap #909
• [Deprecate] DataLoader.linelist() and LineListData class for a while because it is difficult to get regularly updated linelist #866
• [Deprecate] JHUData.from_dataframe() and JHUData.replace() #905
• [Deprecate] DataLoader.population() and PopulationData class #904
• [Deprecate] PCRData.replace() #912
• [Deprecate] CountryData class #913

Phase editting

• [Fix] ValueError when adding phases manually to fully-complemented data #878
• [Fix] KeyError when adding new phases after clearing all past phase #879
• [Fix] results of parameter estimation is not cleared by Scenario.clear(include_past=True) #880

ODE parameter estimation

• [Fix] snl.estimate_accuracy() showing error without snl.trend() #895
• [Fix] Scenario.estimate(): argument phases does not work #911

Forecasting

• [New] [forecasting] include the number of elapsed days from indicator change points in X #842
• [New] [forecasting] user interface to add features to X #843
• [Revise] [forecasting] mask highly correlated features #858
• [Revise] [forecasting] include log-transformed indicator values in X #860

Documentation and discussion

• [Question/Kaggle] SEWIRF Linelist data kde plot #838
• [Docs] update Functionalities section of README #850
• [Docs] how to export high resolution TIFF images #874
• [Discuss] evaluation tool of impact of interventions including vaccination #644

Acknowledgement

Special thanks for your discussions!

• @Inglezos
• @subi10
• @AnujTiwari
• @rebeccadavidsson

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/38?closed=1

CovsirPhy v2.21.0

Data visualization

• [New] class/function for scatter plot #819

ODE parameter estimation

• [Revise] improve parameter estimation performance with shorter timeout_iteration (and constant liar optionally) #833

Forecasting

• [New] select algorithms of forecasting #795
• [New] [forecasting] plot predicted vs. true parameter values #815
• [Revise] [forecasting] use all candidates of delay period to create delayed X #816
• [Revise] [forecasting] use MAPE as the default evaluation metrics and prevent overfitting #813
• [Revise] [forecasting] X include Indicators(n)/Indicators(n-1) #823
• [Revise] [forecasting] hyperparameter tuning of SVR with Bayesian optimization #831

Documentation

• [Docs] update estimate_delay_days.ipynb to fix AttributeError #829
• [Docs] use vaccine data in example/scenario_analysis.py #836

Acknowledgement

@Inglezos and @jparra0419, special thanks for your discussions!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/37?closed=1

CovsirPhy v2.20.3

Installation

• [Fix] error in updating CovsirPhy to 2.20.2 with Anaconda #800

Data loading

• [Fix/Question] Use of local dataset. time not converted to date #791

Forecasting

• [Fix/Revise] [forecasting] data is shuffled when splitting the data to train/test #802
• [Revise] use SVR as a regressor in forecasting #803
• [Fix/Revise] [forecasting] select n_componets of PCA by percentages in decision tree regressor #806

Acknowledgement

@Inglezos and @jparra0419, special thanks for your discussions!

Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/36?closed=1

CovsirPhy v2.20.2

Forecasting

• [Fix] test score is quite low #783
• [Fix] error handling is not clear when a error is raised with regressor #784
• [Fix] low accuracy when backtesting because correlation of ODE parameters are not considered #786
• [Fix] data augmentation improves training/test score just for show #789
• [Discuss] select features independently for all the ODE parameters #778

Acknowledgement

@Inglezos, special thanks for your discussions!
Related issues and pull requests:
https://github.com/lisphilar/covid19-sir/milestone/35?closed=1