Skip to contents

Aggregate individual judgement matrices

Source: R/decision_data.R
sframe_aggregate_judgements.Rd

Aggregation of individual judgements (AIJ) across respondents. The element-wise geometric mean is the standard choice for reciprocal AHP matrices, because it is the only mean that preserves reciprocity: the arithmetic mean of m[a, b] and the arithmetic mean of m[b, a] are not reciprocals of each other. DEMATEL matrices are not reciprocal, so they aggregate arithmetically.

Usage

sframe_aggregate_judgements(
  matrices,
  method = c("geometric", "arithmetic"),
  cr_filter = FALSE
)

Arguments

matrices

Either a list of square numeric matrices or the object returned by sframe_assemble_pairwise().

method

"geometric" (the AHP default) or "arithmetic" (DEMATEL).

cr_filter

Logical. When TRUE, individual matrices with a consistency ratio at or above 0.10 are dropped before aggregation. Applies to reciprocal matrices only. Default FALSE.

Value

A list with matrix, method, n_respondents, n_dropped, n_dropped_consistency, and consistency (the per-respondent CR distribution, or NULL for a non-reciprocal set).

See also

sframe_assemble_pairwise()

Examples

m1 <- matrix(c(1, 3, 1 / 3, 1), 2, 2)
m2 <- matrix(c(1, 5, 1 / 5, 1), 2, 2)
sframe_aggregate_judgements(list(m1, m2))$matrix
#>          [,1]      [,2]
#> [1,] 1.000000 0.2581989
#> [2,] 3.872983 1.0000000