My colleagues Girish Gupta, Vaidehi Tandel and I showcase how and why projections on COVID-19 cases using CDGR need to be mindful of the inherent nuances of using this metric.
Excerpts below: “1. CDGR is highly sensitive to the time period used. The MoHFW estimations used a longer time period of 30 days as compared to our projections — based on a time period of 24 days — which resulted in a very high projection of 650,000. To illustrate, if we use a time period of 30 days starting from 23rd April and 3,683 cases, the projected number of cases will be 520,000 using a CDGR of 18%.
2. The use of a very high CDGR of 20%. This is not borne out by the data on the actual number of confirmed cases over the last eight days in Mumbai: 9%.”
For the full post, click here.