Temporal Persistence

Temporal Persistence measures stability of a metric across time splits (baseline vs target period).

Formula

\[ Persistence = target_score / baseline_score \]

Inputs

• baseline_score
• target_score

Output

• Ratio in [0, 1] in typical cases. Higher is better.

Examples

from tempoeval.metrics import TemporalPersistence

metric = TemporalPersistence()
score = metric.compute(baseline_score=0.8, target_score=0.6)