Sourced fact
The reported result is less than 4% mean absolute error for the study's SoH estimator.
Used EV evidence · UK
What Huo's under-4% estimator error cannot mean
The Huo paper reports less than 4% mean absolute estimation error within its small same-model taxi study. That figure is not battery degradation, annual capacity loss or remaining usable capacity. It also cannot promise the same estimator performance for another vehicle, model, climate or duty cycle.
Reviewed
Key facts and evidence
For UK readers, as of 13 August 2026, Huo's less-than-4% figure is mean absolute error for an estimator validated within a small same-model taxi cohort. It is not a measured degradation rate or a limit on an individual car's error. E2 records no lower or upper uncertainty bound and no external-fleet validation.
Sourced fact
Estimator error must not be relabelled as battery degradation, annual fade or remaining capacity.
Sourced fact
The E2 record has no reported lower or upper bound for this error result.
Sourced fact
Same-fleet validation does not establish estimator performance for a different vehicle population.
The Huo result is an estimator-error measure in one same-model taxi fleet, not a battery degradation or remaining-capacity percentage.
Read the unit first
E2 stores the point result as less than 4 percent mean absolute error for SoH estimation. The metric compares estimated SoH with the study's calibrated reference; it does not state how much capacity the taxis lost.
No interval supplied
The admitted E2 claim has null lower and upper bounds because those values were not available from the source record. Do not invent a range or treat the point result as a worst-case limit.
Source and date
Huo and colleagues published the article on 11 January 2021. E2 retrieved the licensed source on 12 August 2026 and checked the battery-evidence build on 13 August 2026.
Limit and next step
The estimator was trained and validated in one same-model taxi fleet, so transfer performance remains unverified. For an individual car, use E3 row e3-s11-008 and request a report that identifies the vehicle, method, conditions and limitations.
What to do next
Estimator accuracy in a research cohort does not establish the driving condition of a particular vehicle.
Immediate action: Check whether the figure is labelled as estimator error rather than battery loss.
- Reject the interpretation if the error metric is presented as measured degradation or an individual accuracy promise.
Checks that stay with this vehicle
Check that any citation of the Huo figure names mean absolute estimation error, the same-fleet sample and the lack of external validation.
The number remains attached to its estimator metric and narrow study population.
Treat the evidence as misleading if the number is used as degradation, remaining capacity or individual test accuracy.
Professional hand-off
- Ask an EV-competent report issuer to replace any transferred research accuracy claim with vehicle-specific, explained evidence.
Questions this record can answer
What does the sourced evidence say about read the unit first?
E2 stores the point result as less than 4 percent mean absolute error for SoH estimation. The metric compares estimated SoH with the study's calibrated reference; it does not state how much capacity the taxis lost.
What does the sourced evidence say about no interval supplied?
The admitted E2 claim has null lower and upper bounds because those values were not available from the source record. Do not invent a range or treat the point result as a worst-case limit.
What does the sourced evidence say about source and date?
Huo and colleagues published the article on 11 January 2021. E2 retrieved the licensed source on 12 August 2026 and checked the battery-evidence build on 13 August 2026.