Logistic Models of COVID19 in Vermont Over Time
- Author:
- Greg Petrics
- Topic:
- Statistics
As in most of the world, Vermont (VT) is experiencing rapid growth in the number of COVID19 cases. The website VTDigger.org has been publishing the number of cases in Vermont daily.
The dynamic graphic below depicts the following:
- The number of documented cases of COVID19 in VT on every day since March 8, 2020 when the first case was reported. The data is in the table on the left, and plotted as red dots on the right. The data is courtesy of VTDigger.org and the Vermont Department of Health.
- A least-squares logistic growth model of the number of documented cases of COVID19 in VT. The model is plotted as the red curve on the right. You can view changes to the model as more cases have been reported with the slider "Adjust Models".
- The prediction of the logistic growth model for May 1, 2020. The prediction is plotted as a dark blue point. The spread of the predictions is recorded as light blue dots when you "Adjust Models".
- The predicted inflection point of the growth of the number of COVID19 cases in VT. The prediction is plotted as a vertical black bar. The spread of the predictions is recorded as light gray lines when you "Adjust Models".
This graphic will be updated when more data becomes available.
Notes:
- This model should not be used as a decision making tool. The purpose of this document/webpage is educational only, and is meant to illustrate the mechanism, features and drawbacks of least squares logistic regression.
- This model does NOT take into account the number of tests being done. Future documented cases will depend not only on the actual number of cases, but also on the number of tests being done. As such, when the rate or testing changes, the number of documented changes will respond in a way that this model does NOT predict.
- This model does not take into account social distancing measures. It only depends on the number of documented cases. As social distancing measures change in the future, the number of cases, the number of tests, and therefore the number of documented cases will react accordingly in a way that this model does NOT predict.
- That said, this set of logistic models still provides an interesting insight into how logistic models react to the data they're given. Logistic models are very sensitive to their most recent data point, so it is likely that both the inflection point and the carrying capacity will change dramatically as more data is added.