Estimate multi-aircraft frames

A single DataFrame can mix several aircraft typecodes. FuelEstimator.estimate() groups rows by typecode and scores each group with its own aircraft parameters, then scatters the results back into the original row order — so you can process a whole fleet in one call.

A mixed frame

import pandas as pd
from acropole import FuelEstimator

fe = FuelEstimator()

flight = pd.DataFrame({
    "typecode": ["A320", "A320", "B738", "B738"],
    "groundspeed": [400, 410, 420, 430],
    "altitude": [10000, 11000, 12000, 13000],
    "vertical_rate": [2000, 1500, 1000, 500],
})

result = fe.estimate(flight)
# A320 rows scored with A320 params, B738 rows with B738 params

Unsupported typecodes

If a typecode is not present in the bundled aircraft parameters, those rows are left as NaN in fuel_flow and a warning is emitted. Missing typecodes follow the same policy. Without second, known aircraft rows are scored independently:

import warnings

flight = pd.DataFrame({
    "typecode": ["A320", "ZZZZ"],
    "groundspeed": [400, 410],
    "altitude": [10000, 11000],
    "vertical_rate": [2000, 1500],
})

with warnings.catch_warnings(record=True) as caught:
    warnings.simplefilter("always")
    result = fe.estimate(flight)

print(result["fuel_flow"])      # second row is NaN
print([str(w.message) for w in caught])  # "Aircraft type 'ZZZZ' not supported"

With second, each typecode needs enough samples to derive its time-dependent features. A single-sample group has undefined computed derivatives. Any unknown fuel contribution also makes that and all subsequent fuel_cumsum totals unknown. See Missing data and cumulative fuel for the complete policy.

Sharing the session across aircraft

When you repeatedly score one aircraft at a time, use for_aircraft() to obtain a typecode-bound AircraftFuelEstimator that reuses the already-loaded ONNX session and parameters (no reload per typecode):

fe = FuelEstimator()
a320 = fe.for_aircraft("A320")
b738 = fe.for_aircraft("B738")  # same session, different params

See the Reference for AircraftFuelEstimator's numpy-only estimate() signature.