Skip to contents

Create an Uncertainty Component

Source: R/uncertainty.R
uncertainty_component.Rd

Defines a single uncertainty component for use in an uncertainty budget. This follows ISO GUM terminology with Type A (statistical) and Type B (other means) uncertainty evaluation.

Usage

uncertainty_component(
  name,
  value,
  type = c("A", "B"),
  sensitivity = 1,
  df = Inf,
  distribution = c("normal", "rectangular", "triangular", "u-shaped"),
  coverage_factor = 1
)

Arguments

name

Name/description of the uncertainty source.

value

Standard uncertainty value (u).

type

Type of evaluation:

  • "A": Statistical evaluation (from repeated measurements)

  • "B": Evaluated by other means (from specifications, certificates, etc.)

sensitivity

Sensitivity coefficient (c). Default is 1. The contribution to combined uncertainty is |c| * u.

df

Degrees of freedom for this component. Default is Inf (for Type B with no DOF information).

distribution

Distribution assumed for Type B:

  • "normal": Normal distribution (default for Type A)

  • "rectangular": Uniform distribution (common for Type B)

  • "triangular": Triangular distribution

  • "u-shaped": U-shaped distribution

coverage_factor

Coverage factor (k) used to derive this value from an expanded uncertainty. Default is 1 (value is already standard uncertainty).

Value

An uncertainty_component object.

Details

Type A Evaluation

For Type A components, the standard uncertainty is typically the standard error of the mean: u = s / sqrt(n), with df = n - 1.

Type B Evaluation

For Type B components from expanded uncertainties with coverage k: u = U / k. For rectangular distributions: u = a / sqrt(3).

See also

measure_uncertainty_budget() for combining components, measure_uncertainty() for quick uncertainty calculation.

Examples

# Type A: Repeatability from 10 measurements
u_repeat <- uncertainty_component(
  name = "Repeatability",
  value = 0.05,  # Standard error of mean
  type = "A",
  df = 9
)

# Type B: Calibrator uncertainty from certificate (k=2)
u_cal <- uncertainty_component(
  name = "Calibrator",
  value = 0.02 / 2,  # Divide expanded uncertainty by k
  type = "B",
  df = 50
)

# Type B: Temperature effect (rectangular distribution)
u_temp <- uncertainty_component(
  name = "Temperature",
  value = 0.1 / sqrt(3),  # Half-width / sqrt(3) for rectangular
  type = "B",
  distribution = "rectangular"
)