Get the set of weights for building the composite indicator

get_weights returns the weights for constructing the composite indicator.

Usage

get_weights(data, method, expert_weights = NULL, ...)

Arguments

data: data matrix of binary elementary indicators (without missing values).
method: method for getting the set of weights. Possible choices are: "equal", "experts" or "irt". See Details.
expert_weights: user-provided expert weights (if any).
...: optional arguments for mirt::mirt() function. See Details.

Value

vector of weights.

Details

This function returns a vector of weights, whose dimension is equal to the number of elementary indicators in data (say, \(Q\)). Three choices are offered:

"equal": each elementary indicator receives the same weight, equal to \(1/Q\);
"experts": each elementary indicator receives a specific weight provided by experts. The user can provide the weights provided by his/her own experts (through argument expert_weights), otherwise the weights provided by our pool of experts is used;
"irt": each elementary indicator receives a specific weights provided by the IRT framework.

Specifically, a unidimensional 2PL IRT model is estimated on data through mirt::mirt() function, for which possible additional arguments can be provided (...), such as estimation algorithm, numerical optimiser, convergence threshold, etc. Once the model is fitted, weights are computed by rescaling the estimated discrimination parameters (in such a way they sum to 1). Note: for the moment, IRT weights can be obtained only when the elementary indicators are all binary.

Examples

if (FALSE) {
if (interactive()) {
  # sample of 100k contracts
  set.seed(12345)
  i <- sample(1:nrow(mock_data_core), size = 1e5)
  mock_sample0 <- mock_data_core[sort(i), ]

  # indicators for companies
  mock_sample <- tidyr::unnest(mock_sample0, aggiudicatari, keep_empty = TRUE)
  mock_sample_variants <- tidyr::unnest(mock_sample, varianti, keep_empty = TRUE)

  out_companies <- ind_all(
    data = mock_sample,
    data_ind8 = mock_sample_variants,
    emergency_name = "coronavirus",
    target_unit = "companies"
  )
  indicator_data_matrix <- create_indicator_matrix(out_companies)
  w <- get_weights(indicator_data_matrix, method = "equal")
  w
  w <- get_weights(indicator_data_matrix, method = "experts")
  w
}
}