[bergirstmair]

Some questions related to the scoring, maybe this short fictive example helps to clarify.

ExpID | Correct state | Predicted state
1 | healthy | healthy
2 | dzs_1r | dzs_1l
3 | dzp_1r+dzp_2r | dzp_1r+dzp_1l
4 | dzp_1r | healthy

For the first objective, the correct prediction of healthy and faulty regimes of operation of a car will be scored using the accuracy rate:
3 regimes/experiments out of 4 are predicted correct, accuracy = 0.75

Some questions related to the scoring, maybe this short fictive example helps to clarify.

ExpID | Correct state | Predicted state
1 | healthy | healthy
2 | dzs_1r | dzs_1l
3 | dzp_1r+dzp_2r | dzp_1r+dzp_1l
4 | dzp_1r | healthy

For the first objective, the correct prediction of healthy and faulty regimes of operation of a car will be scored using the accuracy rate:
3 regimes/experiments out of 4 are predicted correct, accuracy = 0.75

For the second objective, the sensitivity in detecting specific faults will be measured by the fraction of the sum of correct individual fault type predictions (for all fault types) over the total faulty instances
Total number of faults = 0+1+2+1=4, correct individual fault type predictions =1(third ExpID dzp_1r), sensitivity=0.25

[bergirstmair]

An additional comment on the scoring:

If the correct state has one fault but two faults are predicted the scoring for this case is 0.5

[Author]

Posts