covid19model-fr-regions-results

This dataset is a derivative work from the Imperial College Study Estimating the number of infections and the impact of nonpharmaceutical interventions (NPI) on COVID-19 in 11 European countries. This repository presents results for the forecasting of the COVID-19 epidemic at French regional level.

The goal of this dataset is to open the access to the result of our simulations, we encourage:

• Suggesting runs, to validate specific behaviours.
• Making visualisation of the data that illustrate reliability.
• Analysis of the performance and forecasting.

This dataset is part of the data against covid-19 citizens' initiative for open data and open source code around the COVID-19 pandemic.

Looking to contribute? Check the contributing section below on ways you can help and then go to our projects page!

Codes to run the simulations in our sister repository.

This repository is also available on Kaggle

Licenses

• Data (all .csv and .png files) in this repository are licensed under the CC-BY-4.0 license.
• Code in this repository (all files except .csv and .png) in this repository are licensed under the MIT License.

Using this data (in development)

Analysis of the data is facilitated by an example jupyter notebook which shows usage of the model_analysis python package which is under development.

See the data_example notebook or the model_analysis package for more info.

Example dataset and visualisation

• Tutorial and explained on how to load data and plot itdata_example
• A short notebook analysing the amount of data after a NPI to ensure convergence analysis_report_convergence

Adding to the data

• Fork this repository

• Clone this repository

• Install the git hooks by running:

  .githooks/copy-hooks.sh

• Add the content of the payoto:covid19model/results/ folder corresponding to successful --fullrun executions.

• Do a PR to add to this repository.

Structure of data

• Data is stored in the runs/ folder
• Each sub-directory is a "full" run (ran with flag --full: 4 chains with 4000 iterations and 2000 repetitions).
• .csv files describing the inputs are:
  • <run name>-inputs-active_regions_ifr.csv : List of regions considered and the country they are part of, and the IFR used.
  • <run name>-base-intervention.csv : List of NPIs per region considered.
  • <run name>-inputs-distribution-parameters.csv : Configuration of distributions governing transition from infection to hospital and death.
• Data files describing the finale fitted parameters:
  • <run name>-covars-alpha-reduction.csv : Reduction of the Rt as a result of each considered NPI (see .png file for visualisation).
  • <run name>-final-rt.csv : Final values of Rt for each simulation in each region (see .png file for visualisation).
  • <run name>-final-mu.csv : Final deviation of Rt for each simulation in each region (see .png file for visualisation).
• .csv files describing the modelled time series and input data:
  • <run name>-base-plot.csv : Time series of the fitted data from start time-to simulation time, no forecast.
  • <run name>-forecast-data.csv : 7 days time series forecast for each region.

Detailed description of data files

In progress

-base-plot.csv - model fit time series

• "" : index
• "time" : Date in format yyyy-mm-dd.
• "country" : Country.
• "region" : Geographical region in country.

Other columns are detailed in tabular form below. Notes:

• <XXX>_c : denotes a quantity cumulated since the start of the local epidemic;
• modelled data is presented with confidence intervals:
  • <XXX>_min : denotes the lower bound of a 95% confidence interval (0.025% band);
  • <XXX>_max : denotes the upper bound of a 95% confidence interval (0.975% band);

Column name	source	temporality	Confidence Band	Description
"estimated_deaths_forecast"	forecast	daily	50%	deaths of COVID-19.
"estimated_deaths_forecast_min"	forecast	daily	0.025%	deaths of COVID-19.
"estimated_deaths_forecast_max"	forecast	daily	0.975%	deaths of COVID-19.
"rt"	Inferred	instantaneous	50%	R_t total reproduction number
"rt_min"	Inferred	instantaneous	0.025%	R_t total reproduction number
"rt_max"	Inferred	instantaneous	0.975%	R_t total reproduction number

-forecast-data.csv - model forecast time series

• "" : index
• "time" : Date in format yyyy-mm-dd.
• "country" : The geographical region of the data, to find the country crossreference with <case name>-inputs-active_regions.csv.
• "region" : Geographical region in country.

Other columns are detailed in tabular form below. Notes:

• <XXX>_c : denotes a quantity cumulated since the start of the local epidemic;
• modelled data is presented with confidence intervals:
  • <XXX>_min : denotes the lower bound of a 95% confidence interval (0.025% band);
  • <XXX>_max : denotes the upper bound of a 95% confidence interval (0.975% band);
  • <XXX>_min2 : denotes the lower bound of a 50% confidence interval (0.25% band);
  • <XXX>_max2 : denotes the upper bound of a 50% confidence interval (0.75% band);

Column name	source	temporality	Confidence Band	Description
"reported_cases"	Official	daily	real sample	cases of COVID-19.
"reported_cases_c"	Official	cumulated	real sample	cases of COVID-19.
"predicted_cases_c"	modeled	cumulated	50%	cases of COVID-19.
"predicted_min_c"	modeled	cumulated	0.025%	cases of COVID-19.
"predicted_max_c"	modeled	cumulated	0.975%	cases of COVID-19.
"predicted_cases"	modeled	daily	50%	cases of COVID-19.
"predicted_min"	modeled	daily	0.025%	cases of COVID-19.
"predicted_max"	modeled	daily	0.975%	cases of COVID-19.
"predicted_min2"	modeled	daily	0.25%	cases of COVID-19.
"predicted_max2"	modeled	daily	0.75%	cases of COVID-19.
"deaths"	Official	daily	real sample	deaths of COVID-19.
"deaths_c"	Official	cumulated	real sample	deaths of COVID-19.
"estimated_deaths_c"	modeled	cumulated	50%	deaths of COVID-19.
"death_min_c"	modeled	cumulated	0.025%	deaths of COVID-19.
"death_max_c"	modeled	cumulated	0.975%	deaths of COVID-19.
"estimated_deaths"	modeled	daily	50%	deaths of COVID-19.
"death_min"	modeled	daily	0.025%	deaths of COVID-19.
"death_max"	modeled	daily	0.975%	deaths of COVID-19.
"death_min2"	modeled	daily	0.25%	deaths of COVID-19.
"death_max2"	modeled	daily	0.75%	deaths of COVID-19.
"rt"	Inferred	instantaneous	50%	R_t total reproduction number
"rt_min"	Inferred	instantaneous	0.025%	R_t total reproduction number
"rt_max"	Inferred	instantaneous	0.975%	R_t total reproduction number
"rt_min2"	Inferred	instantaneous	0.25%	R_t total reproduction number
"rt_max2"	Inferred	instantaneous	0.75%	R_t total reproduction number