Average label uniqueness

q.label.average_uniqueness averages inverse concurrency over each event lifetime. An isolated label has uniqueness 1; overlapping labels receive smaller values.

Use uniqueness as a diagnostic, a decay input, or one component of explicit training weights.

import pandas as pd

import qrt as q

observations = pd.date_range("2026-01-01", periods=7, name="datetime")
end_times = pd.Series(
    observations[[3, 5, 6]],
    index=observations[[0, 1, 4]].rename("event_time"),
)
uniqueness = q.label.average_uniqueness(observations, end_times)
pd.DataFrame({"end_time": end_times, "average_uniqueness": uniqueness})

	end_time	average_uniqueness
event_time
2026-01-01	2026-01-04	0.625000
2026-01-02	2026-01-06	0.500000
2026-01-05	2026-01-07	0.666667

Precomputed concurrency can be supplied when several analyses share the same intervals. It may be zero outside all event lifetimes but must be positive inside every event.