Saw these shared on social media today.
This is the model from last December projecting daily new cases under various scenarios. The column chart at bottom are the actual real world numbers that occurred. Since they didn’t know which scenario would happen, their range went from nothing to infinity.
The next chart is projected deaths due to Covid-19, from mid-January (solid red line). The actual real world deaths are plotted in the individual red points, below, and even well below the 95% confidence band.
This table compares the March 2020 Imperial College London (ICL, Neil Ferguson) projected deaths with actual deaths. Note – the actual deaths in Japan were over 9,000 at end of March – the value of 10 is an error in this table.
Disease models appear to be unfit for any purpose. We have seen this before.
Oregon’s disease model – projecting only 3 weeks into the future, was completely wrong 11 out of 13 times.