Community contribution: Python analysis and plotting tools for output CSVs

[payoto]

This is not really an issue and more of an announcement/sharing of some derivative work that has been going on in a the French citizen's initiative data against covid. Some of the codes and tools developed may be of interest to the original authors and/or other people trying to exploit the model with for different countries/with different data.

First, possibly of most interest to the community, we have been developing a python package for reading, visualising and analysing the results of this model in python, using pandas and matplotlib. Below are a couple of example usages in jupyter notebooks:

• Explainer notebook
• Analysis of the number of days of data required for the model to converge on an effect of NPI

In our effort to model the progression of the epidemic at finer geographical resolution in France we've added some additional flexibility to the input parameters. You can find all of this in my fork of the project payoto/covid19model.

In order to validate these changes and understand the predictive capability of the model we have also been collecting full simulation runs with data from different dates. This collection effort is stored at https://github.com/payoto/covid19model-fr-regions-results and is also available on kaggle. This data is used to validate the model and is made available for analysis under permissive licences.

Thanks for your hard work on updating this model!

[payoto]

Our version of the three panel plot (length of forecast for validation only):

These can then be assembled in markdown documents in few lines of code:

Automatically generated country reports