Anvil.Agreement.Fleiss (Anvil v0.1.1)

View Source

Fleiss' kappa for measuring agreement among multiple raters.

κ = (P̄ - P̄_e) / (1 - P̄_e)

where:

  • P̄ = mean observed agreement across samples
  • P̄_e = expected agreement by chance

Summary

Functions

compute(labels, opts \\ [])

Computes Fleiss' kappa for n raters.

Functions

compute(labels, opts \\ [])

@spec compute(
  [Anvil.Label.t()],
  keyword()
) :: {:ok, float()} | {:error, term()}

Computes Fleiss' kappa for n raters.

Options

  • :field - Field name to compute agreement for (default: first field in values)