Metric
WindowedF1Score
F1-score for event detection, using a margin-based match criterion.
This score computes an iloc-based F1-score with margin of errog
A true event is considered a match if there is a predicted event within a margin of error, as specified by the margin parameter.
The margin is applied in iloc units, i.e., the absolute difference between the true and predicted iloc must be less or equal than the margin to be considered a match.
The default value of 0 results in only exact matches being counted.
Schnellstart
python
from sktime.performance_metrics.detection import WindowedF1Score
estimator = WindowedF1Score(margin=0)Parameter(1)
- marginint, optional (default=0)
- Margin of error to consider a detected event as matched.