I have a final datapoint from our 2015 e-Golf. We extended the lease to bridge the time until we got our Model 3, but we have it now, so the e-Golf is going back on Monday.

As I indicated before in this thread, I think this method is significantly flawed. However, I tried a couple times to take different data on consecutive days to highlight the variability in the results so people can judge for themselves.

As you can see from the data above, the least squares fit line gives an estimated initial usable capacity of just under 22kWh and a degradation rate of 2.09Wh/day with a final capacity at the end of the trend line of 19.46kWh. However, the last actual data point was calculated as 20.0kWh. The calculated degradation slope correlates to 3.5% degradation per year.

This scatter chart shows how the amount of battery used in the calculation contributes to the resulting value. The yellow and red lines represent data that is time and mileage adjacent, so there's definitely a correlation there.

This last chart shows how the first data point from October 2015 skews the results. When that point is removed and least squares fit is performed on the remaining data, the slope of the degradation line is reduced to 1.13Wh/day and the estimated original usable capacity is lowered to 21.01kWh. That comes to less than 2% degradation per year.

So, that's my data for from our car. It would be much better if we could get more accurate capacity data from the car's computers, but this methodology is the best I could come up with in its absence. The VIN ends in 907055 if anyone cares to look for it later. It's a Limestone Gray 2015 e-Golf LE.