hyfe_performance: Evaluate Hyfe's performance based on a set of labelled...
In hyfe-ai/hyfer: R Utilities for Interacting with Hyfe Data

Evaluate Hyfe's performance based on a set of labelled sounds.

hyfe_performance(
  detections,
  labels,
  offsets,
  offset_cutoff = 3,
  prediction_threshold = 0.7,
  n3_cutoff = 10,
  quality_cutoff = 0.95,
  remove_tests = NULL,
  toplot = TRUE,
  verbose = TRUE
)

`detections`	A `dataframe` of Hyfe detections. This dataframe requires these columns: `alias` (research/device ID); `timestamp` (with a numeric timestamp); `n_peaks` (the number of peaks occurring at the time represented by the timestamp; this can just be 1 if you don't know); and `prediction_score` (0-1).
`labels`	A `dataframe` with analyst labels. Required columns: `test` (numeric ID for each test labeled); `timestamp` (numeric timestamp for the cough-second); `label` (best label for this cough-second, 1 - 3, using the Hyfe 4-tier system: 1 = a disputable cough, possibly fake; 2 = definitley a cough but barely audible; 3 = a clear authentic cough); `n1` (if this cough second was labeled category 1 by any analysts, this value is 1; if not it is 0); same for `n2` (category 2) and `n3` (category 3); `n_labels` provides the number of labels that occurred within this cough-second; `n_analysts` provides the number of analysts who have reviewed this test; `valid` indicates whether this cough-second should be included in the analysis (`TRUE` or `FALSE`).
`offsets`	A `dataframe` of time offsets, in seconds, for the Hyfe detections in each test represented in the `labels` dataset. Each row is a test. Required fields are: `alias` (research/device ID); `test` (numeric ID for the test); `offset` (seconds by which to adjust Hyfe timestamp; if postivie, Hyfe timestamps will increase; if negative, they will decrease).
`offset_cutoff`	A number indicating the allowable difference, in seconds, between a Hyfe detection and a label timestamp.
`prediction_threshold`	Cough prediction score threshold.
`n3_cutoff`	Minimum number of category 3 coughs that must be labeled for a test in order for that test to be included in the calculation of performance metrics.
`quality_cutoff`	Minimum quality of a test in order for that test to be included in the estimation of performance metrics. Quality is defined as the fraction of cough-seconds that are category 3.
`remove_tests`	Numeric vector providing the IDs for any tests you want to remove from the performance analysis.
`toplot`	Plot diagnostic plots?
`verbose`	Plot status updates?