Boom. Today was the largest day for new COVID-19 infections: 269 in King County.
However, due to the changes in how King County Health is reporting data, only 83 of these are attributed to “yesterday.” The rest are spread pretty evenly across the past week. for example:
- 3/28: +18
- 3/29: +22
- 3/30: +17
- You get the idea
What this isn’t great for is building a robust model: the data keeps changing and historical numbers are pushed up, but recent numbers are artificially low, likely because it takes time to turn around tests.
The county also stopped reporting total number of test administered and negative test results. Thanks for nothing!
The model built with 2 weeks of data no longer has any predictive power. The r-squared dropped precipitously to 0.0002. Yikes.
The daily model continues to plug along, with an r-squared of 0.836 and predicting 189 for tomorrow. The progress from the past few days has been completely erased, and now the model predicts we’ll continue to gain 5+ infections from the previous day.