It is not accurate measurement but looking at percentages it gives you a rough idea.
+1 to this.
If EDSM originally was tracking 10k commanders, and now it says only 6k are left, then you can estimate that there is approximately 60% of the entire group left. So, if 15k people were in the starting set, then it is not an unreasonable estimate to guess that there are about 9k total left. Note, the numbers I give here are just examples, not actual numbers.
This is not to say that there are not systemic sources of bias in this estimate. For example, perhaps people who are logging data to EDSM are more "hardcore", and therefor less likely to drop off then the group of people who are not logging data. If this was the case, then perhaps the attrition rate for non-logging group would be higher than 40%.
In the absence of hard data to provide a quantitative correction for sources of bias, however, the best one can do is present the numbers we do know about (the starting number of tracked individuals and the remaining number of tracked individuals), and then extrapolate. It is important to always keep in mind that the uncorrected extrapolation is, at best, a rough guess (as said by Dr. Nagi)