Compute outlier flags using sliding window.

Same as identify_vector_outliers_global() except the quantile computation is performed in a sliding window centered on each sample. Currently the computation is not really efficient and runs in O (n*w*log (w)) for window size w.

Usage

identify_vector_outliers_window(
  .input,
  .window = 333,
  .limit = 0.05,
  .slack = 0.1
)

Arguments

.input

Sample vector.

.window

Window size in samples.

.limit

Quantile that separates inliers from outliers.

.slack

Tolerated distance from limit quantile.

Value

Boolean vector of outlier flags.

See also

identify_vector_outliers_global()

Examples

# A sample of 100 is considered an outlier when near
# different values but not when near similar values.
identify_vector_outliers_window (c (100, 15:25, 95:105, 100), .window = 10)
#>  [1]  TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
#> [13] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE