My goal is to find the best performing forecasting model for the occurrence of COVID-19 in Toronto. I pre-train the network with data on the occurrence of SARS in ten countries and Toronto. Then I train the network with the data on the occurrence of COVID-19 in the same regions and factors which can affect this number (like age, population density, etc.). How accurate would it be to use this model to predict occurrence of COVID-19 and the number of deaths caused by it in Toronto? Thank you for your help.

  • $\begingroup$ Imho at this stage there's enough data about COVID without relying on SARS data. There's a big risk that the "translation" from SARS to COVID would not work as expected. $\endgroup$ – Erwan Jun 23 at 16:38

So the question is about whether you could use the model you have described to create valid predictions of COVID-19 occurrence in Toronto.

I would say this is depend on the data you have used. Check the distribution of features (age, population density, etc.). If there is large variation in the features(such that the Toronto values are observed as outliers against the others), this could lead to large variation and skew in the model's predictions.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.