Summarises the distribution and effective information content of a set of precision weights.
Value
Named list:
- n_obs
Total observations (including NA).
- n_valid
Observations with finite, positive weights.
- n_eff
Effective sample size: \([\sum w_i]^2 / \sum w_i^2\). Equals n_valid when all weights are equal; decreases as weights become more heterogeneous.
- eff_ratio
n_eff / n_valid. Ranges from 0 to 1; 1 = uniform.
- weight_ratio
max(w) / min(w) among valid weights.
- gini
Gini coefficient of weights. 0 = perfectly uniform, approaching 1 = highly concentrated.
Examples
w <- c(1.5, 1.2, 0.8, 0.3, 0.1)
weight_diagnostics(w)
#> $n_obs
#> [1] 5
#>
#> $n_valid
#> [1] 5
#>
#> $n_eff
#> [1] 3.4
#>
#> $eff_ratio
#> [1] 0.687
#>
#> $weight_ratio
#> [1] 15
#>
#> $gini
#> [1] 0.379
#>