independent_length_sampler {Xcertainty}R Documentation

MCMC sampler for individuals with independent measurements.

Description

Build an MCMC sampler that uses calibration data to estimate independent, unknown measurements. This model assumes all Subject/Measurement/Timepoint combinations are independent. So, this sample is well suited for data containing individuals that either have no replicate samples or have replicate samples that are independent over time, such as body condition which can increase or decrease over time, as opposed to length which should be stable or increase over time. It can also be used to estimate lengths when there are replicate measurements. However, since the model assumes all Subject/Measurement/Timepoint combinations are independent, no strength will be borrowed across temporal replication of a subject's measurements, for example.

Usage

independent_length_sampler(data, priors, package_only = FALSE)

Arguments

data

Photogrammetric data formatted for Xcertainty models, required to be an object with class obs.parsed, which can be obtained by running parse_observations()

priors

list with components that define the model's prior distribution. See help("flatten_data") for more details.

package_only

TRUE to return the formatted data used to build the sampler, otherwise FALSE to return the sampler

Value

outputs a function to run a sampler, the function arguments are:

niter

set the number of iterations

burn

set the number samples to discard

thin

set the thinning rate

Examples

# load example wide-format data
data("calibration")
data("whales")
data("whale_info")

# parse calibration study
calibration_data = parse_observations(
  x = calibration, 
  subject_col = 'CO.ID',
  meas_col = 'Lpix', 
  tlen_col = 'CO.L', 
  image_col = 'image', 
  barometer_col = 'Baro_Alt',
  laser_col = 'Laser_Alt', 
  flen_col = 'Focal_Length', 
  iwidth_col = 'Iw', 
  swidth_col = 'Sw',
  uas_col = 'uas'
)

# parse field study
whale_data = parse_observations(
  x = whales, 
  subject_col = 'whale_ID',
  meas_col = 'TL.pix', 
  image_col = 'Image', 
  barometer_col = 'AltitudeBarometer',
  laser_col = 'AltitudeLaser', 
  flen_col = 'FocalLength', 
  iwidth_col = 'ImageWidth', 
  swidth_col = 'SensorWidth', 
  uas_col = 'UAS',
  timepoint_col = 'year'
)

# build sampler
sampler_data = independent_length_sampler(
  data = combine_observations(calibration_data, whale_data),
  priors = list(
    image_altitude = c(min = 0.1, max = 130),
    altimeter_bias = rbind(
      data.frame(altimeter = 'Barometer', mean = 0, sd = 1e2),
      data.frame(altimeter = 'Laser', mean = 0, sd = 1e2)
    ),
    altimeter_variance = rbind(
      data.frame(altimeter = 'Barometer', shape = .01, rate = .01),
      data.frame(altimeter = 'Laser', shape = .01, rate = .01)
    ),
    altimeter_scaling = rbind(
      data.frame(altimeter = 'Barometer', mean = 1, sd = 1e1),
      data.frame(altimeter = 'Laser', mean = 1, sd = 1e1)
    ),
    pixel_variance = c(shape = .01, rate = .01),
    object_lengths = c(min = .01, max = 20)
  ),
  # set to false to return sampler function
  package_only = TRUE
)

[Package Xcertainty version 1.0.0 Index]