Skip to contents

Filter the data

Source: R/denoiser.R
denoiser-function.Rd

The denoiser() function takes in a dataset and attempts to filter out the inherent measurement error. The function makes use of two steps: The data will go through a Kalman filter and will then be binned, both according to the specifications of the user (see kalman_filter() and bin()). Note that only the first step is mandatory and that it's left up to the user whether they would also like to bin their data through specifying the argument binned.

Usage

denoiser(
  data,
  cols = NULL,
  .by = NULL,
  binned = FALSE,
  span = 0.5,
  fx = mean,
  ...
)

Arguments

data

Dataframe that contains information on location (x- and y-coordinates) and the time at which the measurement was taken. By default, denoiser() will assume that this information is contained within the columns "x", "y", and "time" respectively. If this isn't the case, either change the column names in the data or specify the cols argument.

cols

Named vector or named list containing the relevant column names in data if they do not conform to the prespecified column names "time", "x", and "y". The labels should conform to these prespecified column names and the values given to these locations should contain the corresponding column names in that dataset. Defaults to NULL, therefore assuming the structure explained in data.

.by

String denoting whether the moving window should be taken with respect to a given grouping variable. Defaults to NULL.

binned

Logical denoting whether to also bin the data (TRUE) or to leave the data unbinned (FALSE). Defaults to FALSE.

span

Numeric denoting the size of the bins. Will pertain to the values in the "time" variable. Defaults to 0.5. Ignored when binned is FALSE.

fx

Function to execute on the data that falls within the bin. Will be executed on the "x" and "y" columns separately and should ouput only a single value. Defaults to the function mean(). Ignored when binned is FALSE.

...

Additional arguments defining the Kalman filter to employ for filtering. See kalman_filter().

Value

Smoothed and/or binned data.frame with a similar structure as data

See also

bin() kalman_filter() noiser()

Examples

# Generate data for illustration purposes. Movement in circular motion at a
# pace of 1.27m/s with some added noise of SD = 10cm.
angles <- seq(0, 4 * pi, length.out = 100)
coordinates <- 10 * cbind(cos(angles), sin(angles))
coordinates <- coordinates + rnorm(200, mean = 0, sd = 0.1)

data <- data.frame(
  X = coordinates[, 1],
  Y = coordinates[, 2],
  seconds = rep(1:50, times = 2),
  tag = rep(1:2, each = 50)
)

# Use the denoiser function to get rid of the noise. Kalman filter is 
# defined with the constant velocity model and an error variance of 0.01. 
# Binning is performed with a span of 5 seconds and using the mean of the 
# interval as representatitive of the position within that interval.
denoiser(
  data,
  cols = c(
    "time" = "seconds",
    "x" = "X",
    "y" = "Y"
  ),
  .by = "tag",
  model = "constant_velocity",
  error = 0.01,
  binned = TRUE,
  span = 5,
  fx = mean
)
#>    seconds          X         Y tag
#> 1      3.5  9.2507265  3.033118   1
#> 2      9.0  5.1979924  8.409013   1
#> 3     14.0 -0.7761729  9.829786   1
#> 4     19.0 -6.4355866  7.438479   1
#> 5     24.0 -9.6533994  2.183904   1
#> 6     29.0 -9.0387567 -4.048018   1
#> 7     34.0 -4.9286092 -8.491271   1
#> 8     39.0  0.9964281 -9.717818   1
#> 9     44.0  6.6623475 -7.234252   1
#> 10    48.5  9.5726751 -2.484448   1
#> 11     3.5  9.0264648  3.626843   2
#> 12     9.0  4.6791670  8.701344   2
#> 13    14.0 -1.4118083  9.659137   2
#> 14    19.0 -6.8934020  6.974086   2
#> 15    24.0 -9.8324241  1.521422   2
#> 16    29.0 -8.7847143 -4.512127   2
#> 17    34.0 -4.3072172 -8.830867   2
#> 18    39.0  1.7314943 -9.636223   2
#> 19    44.0  7.0648107 -6.827238   2
#> 20    48.5  9.7296863 -1.911672   2