Density-ratio diagnostics from an LMTP fit — margot_lmtp_weight_diag_from

Constructs per-wave density-ratio summaries and cumulative effective sample sizes (ESS) directly from an LMTP model object. For incremental propensity score interventions (IPSI) with $\delta > 1$, zeros in the density ratios arise solely from dropout, so the helper reconstructs the observation mask via `density_ratios > 0`, computes per-wave quantiles/tail probabilities, applies optional right-tail winsorisation, and reports ESS for the cumulative weight trajectories (raw and trimmed).

Usage

margot_lmtp_weight_diag_from_fit(
  fit,
  outcome = NULL,
  shift = NULL,
  trim_right = 0.999,
  thresholds = c(5, 10, 25, 50, 100)
)

Arguments

fit: Either a single LMTP model (anything exposing `$density_ratios`) or a full `margot_lmtp()` result.
outcome: Optional outcome name (required when `fit` is a full `margot_lmtp()` run).
shift: Optional shift name (clean suffix or full). Required when `fit` is a full `margot_lmtp()` run with multiple shifts.
trim_right: Numeric in `(0, 1]`; optional right-tail winsorisation level applied per wave before forming cumulative products (default `0.999`).
thresholds: Numeric vector of ratio thresholds used when computing tail mass (`Pr(r_t > a)`).

Value

A list with `wave_table` (per-wave diagnostics), `mask_from_fit` (logical observation mask), `density_ratios`, and cumulative weights (`w_cum_raw`, `w_cum_trim`).

Examples

dr <- matrix(c(2, 0, 1,
               1.5, 0.5, 2), nrow = 2, byrow = TRUE)
fit <- list(density_ratios = dr)
diagnostics <- margot_lmtp_weight_diag_from_fit(fit)
diagnostics$wave_table
#>   wave n_obs prop_censored r_q50 r_q90 r_q95 r_q99 r_q999 rtrim_q99 rtrim_q999
#> 1    1     2           0.0  1.75  1.95 1.975 1.995 1.9995  1.994505   1.999001
#> 2    2     1           0.5  0.50  0.50 0.500 0.500 0.5000  0.500000   0.500000
#> 3    3     2           0.0  1.50  1.90 1.950 1.990 1.9990  1.989010   1.998001
#>   frac_gt_5 frac_gt_10 frac_gt_25 frac_gt_50 frac_gt_100 ess_cum_raw
#> 1         0          0          0          0           0        1.96
#> 2         0          0          0          0           0        1.00
#> 3         0          0          0          0           0        1.00
#>   ess_cum_trim
#> 1     1.960067
#> 2     1.000000
#> 3     1.000000