Skip to contents

Automatic labelling of a tag

Source: R/tag_label_auto.R
tag_label_auto.Rd

This function uses acceleration data to classify migratory flights. The function uses a k=2 mean clustering (kmeans()) to identify high activity periods. Periods of high activity lasting more than min_duration are then considered to be migratory flight.

Usage

tag_label_auto(
  tag,
  min_duration = 30,
  thr_reclassify = 0.1,
  post_proc_window = 2
)

Arguments

tag

a GeoPressure tag object.

min_duration

Minimal duration (in minutes) to consider a high activity as migratory flight.

thr_reclassify

Post-processing threshold of activity considered for re-classification. Typically between 0 and 0.5 (no effect). See post-processing for details.

post_proc_window

Post-processing windows considered for re-classification. Typically 0 (no effect) to 5 in unit of the temporal resolution. See post-processing for details.

Value

Same data logger list than input tag, but with the column label filled with "flight" in the acceleration data.frame when a sustained high-activity period is detected.

Details

Additionally, we perform a post-processing step to improve the classification of low-activity period happening during a migration flight (e.g. gliding phase):

  1. Mid-activity surrounded by flights. The kmeans classification classify high-activity all values above 50% of the distance between the two clusters. Here, we classify as mid-activity non-flight activity data which value are greater than thr_reclassify of the distance between the two kmeans clusters. Any periods of mid-activity surrounded by flight are also considered as flights.

  2. To avoid stap lasting only a few datapoints, we also classify any datapoint which is surrounded by a flight before and after, within a window of +/- post_proc_window points. As a result, no stap will be shorter than (2 * post_proc_window + 1) * dt.

This function is inspired by the function classify_flap from the PAMlr package.

See also

GeoPressureManual

Other tag_label: tag_label(), tag_label_read(), tag_label_stap(), tag_label_write()

Examples

withr::with_dir(system.file("extdata", package = "GeoPressureR"), {
  tag <- tag_create("18LX", quiet = TRUE)

  tag <- tag_label_auto(tag, min_duration = 15)
  str(tag$acceleration)
})
#> 'data.frame':	4032 obs. of  4 variables:
#>  $ date : POSIXct, format: "2017-07-27 00:00:00" "2017-07-27 00:05:00" ...
#>  $ value: int  0 0 0 0 0 0 0 0 0 0 ...
#>  $ pitch: int  26 27 27 28 28 28 28 27 28 27 ...
#>  $ label: chr  "" "" "" "" ...