Skip to contents

Calculate species-specific confidence thresholds for BirdNET detections

Source: R/birdnet_calc_threshold.R
birdnet_calc_threshold.Rd

Computes species-specific confidence thresholds for BirdNET detections using either a target precision or a target predicted probability from a logistic regression model. The function returns one threshold per species. The precision-based approach follows Tseng et al. (2025), while the probability-based method is adapted from Wood and Kahl (2024).

Usage

birdnet_calc_threshold(
  validated_data,
  full_data = NULL,
  probability = NULL,
  precision = NULL
)

Arguments

validated_data

A data frame of validated BirdNET detections with columns common_name, confidence, and validation. The validation column must contain 1 (true positives) and 0 (false positives).

full_data

Optional. A data frame of all BirdNET detections with at least common_name and confidence columns. If NULL, validated_data is used instead.

probability

Numeric. A target predicted probability (between 0 and 1) used to calculate thresholds from the logistic regression model.

precision

Numeric. A target precision (between 0 and 1) used to select the lowest threshold that achieves the desired precision.

Value

A tibble with two columns:

common_name

Species common name.

threshold

The calculated confidence threshold for that species.

Details

You must supply exactly one of precision or probability. If both or neither are provided, the function will throw an error.

When using the precision method, the function predicts probabilities for each detection using a logistic regression model fit to validated_data. It then identifies the lowest confidence threshold that meets or exceeds the target precision.

When using the probability method, the function calculates the confidence threshold corresponding to the inverse logit of the target predicted probability from the regression model.

All thresholds are clamped to fall between the minimum observed confidence in validated_data and 1.

References

Tseng, S., Hodder, D. P., & Otter, K. A. (2025). Setting BirdNET confidence thresholds: Species-specific vs. universal approaches. Journal of Ornithology. https://doi.org/10.1007/s10336-025-02260-w Wood, C. M., & Kahl, S. (2024). Guidelines for appropriate use of BirdNET scores and other detector outputs. Journal of Ornithology. https://doi.org/10.1007/s10336-024-02144-5