Validating the Cda and Crr estimation (with Vuelta Opening Team Time Trial Data)
One key factor of measuring Cda and Crr is obtaining valid numbers. The problem with live Cda and Crr testing is that there are so many factors to take into account, so it's difficult to say wether or not the estimations obtained are accurate. With the Virtual Elevation Monte Carlo approach there is basically no way of saying how accurate the estimations are. Of course, you can compare the guessed slope against the real one, but that's not a consistent way of testing. Neither does it provide a method of saying that the last x seconds of data/forces correlates with the incoming ones. Therefore, in the last couple of months I have tried to implement the Linear Regression approach to obtain R squared. By checking against the R squared I can determine if the numbers obtained are trustworthy. I can be sure spikes in data or data loss doesn't affect my interpretation of the data. This will also ensure consistency of the Cda and Crr test runs over time. The Linear Regressi