Orr's model ix three Week forward coronavirus case predictions 04 12-2020
1. Orr’s “Model IX” Predictions Of
Unites States Coronavirus Cases
James K. Orr
April 12, 2020
2. Challenging Predictions
• Score Card From Last Week’s Predictions
• The differences between Model VI and VII
were very, very minor changes in how the
Growth In New Case per day (%/Day) changed.
• This week’s new data has presented even
more issues as the change in %/Day became
more random.
4/9/2020 Actual Model VI Model VII
United States 459,989 566,991 467,097
Texas 10,230 11,296 9,437
3. Formulating Approach For Model IX
• To simulate the latest data required introducing
some random variation in the model, hence the
new projections are for multiple independent
runs of the same model allowing different values
of random variables.
• Model XII looked like a good logic solution, but in
doing a back check predicting over the last two
weeks, it ran slight high
• Model IX had best match for total USA new cases
per day over the last two weeks.
4. Ran Out Of Time
• The challenges has consumed my available time
this weekend. Hence, I will put out limited data.
• I will show the following:
– New three week ahead projection
– New Cases Per Day for three week ahead projection
– New Cases Per Day for back check predicting over the
last two weeks
– Plot of New Case Growth Per Day for input data
• Will do for USA, New York, Texas, and
Massachusetts to illustrate Model IX results