I cannot confirm the eccentricity value of a TLE using corresponding position and velocity vectors. Let me go through an example, ISS, to explain the situation.
Using this TLE for ISS,
1 25544U 98067A 17198.89938657 .00000988 00000-0 22167-4 0 9998
2 25544 51.6416 245.2318 0005849 47.2823 302.7554 15.54170925 66526
Tle data says eccentricity is 0.0005849. Propagating this tle for 0 minutes to get TEME vectors yield,
r=-3.468881045420031e+03, -5.752706429352231e+03, -9.339830044476138e+02
v= 3.693246416348322 , -3.17757480606644, 5.92375001851741
Using ordinary orbital mechanics formulations to obtain eccentricity from those vectors finds eccentricity as 0.0016. The error is 168% compared to TLE data.
Then, I tried converting to ECI since an inertial system is required to use usual orbital mechanics formulations. Converting TEME to ECI with a lot of assumptions about polar motion and nutation finds,
r = -3492.98975522683, -5739.01656212581, -928.333286976148
v = 3.69073985425235, -3.19228499069521, 5.91739984184932
eccentricity is again 0.0016.
What is the reason for this big difference in TLE and orbital calculations?
(eccentiricy is found as 0.0750 when calculated with ECEF vectors).
I also found this website (http://www.tle.info/data/ISS_DATA.TXT), it gives information about ISS orbit. First, it lists some orbital properties for ISS. It says eccentricity is 0.0005425, however, on the same page it also lists keplerian elements and says that eccentricity is 0.0018464. They have the same inconsistency that I'm trying to solve.
I'm aware TLE elements and coordinate transformations I used are not very precise but I don't think they would cause this much error.