The next ten days

The next ten days

Looking to the numbers of Coronavirus epidemic so far, we can understand the serious concern pervading the scientific community and leading to the draconian containment measures taken in China, in Italy and soon in many other countries.

We tried to forecast the next ten days of the epidemic at global level, using the dataset from ECDC (link) which collects the daily positives and deaths coming from more than one hundred countries worldwide (starting from December 31, 2019 to March 9, 2020). For our estimates we used a Convolutional Neural Network model specialized in learning temporal sequences by minimizing the loss function on the second derivative for the cumulative time series. We focused our projections on the worldwide dynamics, but, as you may know, every country provides different statistics, measuring in slightly different ways the epidemic: at national level the contagion may appear quite different.

Below you can see our best projections (confidence interval 80%). The next ten days will be a critical turning point to decide the progression of Covid 19.

UPDATE 19/3/2020: Up to the present day ECDC officially reports 213.254 total cases of Covid-19 worldwide. Comparing the actual data to the previous forecast of ten days ago, we can (sadly) assert that the spreading of the new Coronavirus has entered a phase of exponential growth at global level.

No alt text provided for this image


要查看或添加评论,请登录

Giancarlo Vercellino的更多文章

  • Null Hypothesis #10: Wild assumptions, and educated guesses

    Null Hypothesis #10: Wild assumptions, and educated guesses

    Hypothesis "You can tame your wildest guesses—one node at a time". Bayesian Networks are like the ultimate guessing…

  • The debate of the Century

    The debate of the Century

    I asked ChatGPT to imagine a dialogue among Elon Musk, Yuval Harari and Sam Harris in which they are making fun of each…

  • Null Hypothesis #9: Serial creators

    Null Hypothesis #9: Serial creators

    Hypothesis "Achieving serial success in Kickstarter campaign funding cannot be explained with simple performance…

  • Null Hypothesis #8: Crypto-Quakes or Crypto-Apocalypse?

    Null Hypothesis #8: Crypto-Quakes or Crypto-Apocalypse?

    Hypothesis "The odds for a cryptoquake are not so negligible as you can imagine" Quadriga: April 11, 2019. Voyager:…

    1 条评论
  • Null Hypothesis #7: vertical limit

    Null Hypothesis #7: vertical limit

    Hypothesis "Verticalization strategy saved the bottom line of many IT and Software Providers during the pandemic" This…

  • Null Hypothesis #6: hot air vs "hot air"

    Null Hypothesis #6: hot air vs "hot air"

    Hypothesis "Extreme weather events have a limited impact on climate change awareness" The idea that climate change…

  • Null Hypothesis #5: chasing the Jonases

    Null Hypothesis #5: chasing the Jonases

    Hypothesis "Adoption of new products and technologies has accelerated in the last twenty years" The diffusion of new…

  • Null Hypothesis #4: lockdown eyeballs

    Null Hypothesis #4: lockdown eyeballs

    Hypothesis "Lockdowns had a positive impact on Netflix's subscriber base" Well, it may seem quite a trivial question:…

  • Is "Resilience" the right word? An exercise in meme prediction

    Is "Resilience" the right word? An exercise in meme prediction

    The point: can the amplified awareness about Resilience significantly improve the awareness about Cloud? In this…

  • How to sell a Cookbook the smart way

    How to sell a Cookbook the smart way

    I know, at first glance the comparison of Artificial Intelligence services to Cookbooks may seem somehow daring, but…