Show the Community: Developing a prediction model for the weekly recorded COVID-19-related fatalities

hi everyone. As a follow up/sidetrack of our previous analysis I decided to try developing a prediction model for the weekly recorded COVID-19-related fatalities. My objective was to use very simple features (mobility/social distancing and autoregressive features) and project an estimation for one-week-out. Here is the description of the model, its out-of-sample performance and future projections. Obviously there are several ways that this model can be improved, and there are already models that perform pretty well online (no need for more) but I thought that at least the code might be helpful for people wanting to get started on similar approaches: epiDAMIK20-COVID/COVID19-prediction.ipynb at master · kpelechrinis/epiDAMIK20-COVID · GitHub Any comments/suggestions are always welcome.

Hi @Konstantinos_Pelechrinis_University_of_Pittsburgh, are you familiar with the SafeGraph “awesome-list”? It is essentially a list of awesome code the community has come up with! Would you consider adding a link to this great notebook there?


Thanks; most certainly. I will add it

this is super cool