Summarize what a collection of fits actually did

adjustment and fallback_reason report on one fit at a time, which is not usable across a pipeline of hundreds. This walks a list of fits and returns one row each, so a fallback that quietly changed the estimator for a handful of specifications is visible without inspecting every fit by hand.

Usage

fallback_summary(fits, only_fallbacks = FALSE)

Arguments

fits

A list of fits from lm_lin_lasso, lm_robust_lasso, or lm_moderator_lasso. If the list is named, the names are used to label rows. Elements that were not produced by this package are reported with adjustment = NA.

only_fallbacks

Logical. If TRUE, return only the rows where a fallback fired (default FALSE).

Value

A data frame with columns fit (name or index), adjustment, n_selected, selected, and fallback_reason.

Details

A fallback is not an error: it means the requested adjustment could not be produced and a simpler specification was used instead. That is usually the right thing to do, but it changes what the estimate is, so it belongs in the record rather than in the wrapper.

See also

Other fallback reporting: adjustment(), fallback_log()

Examples

set.seed(1)
n <- 400
dat <- data.frame(Z = rep(0:1, n / 2), X_sig = rnorm(n), X_noise = rnorm(n))
dat$Y <- 0.5 * dat$Z + 2 * dat$X_sig + rnorm(n)

fits <- list(
  informative = lm_lin_lasso(Y ~ Z, ~ X_sig + X_noise, data = dat),
  noise_only  = lm_lin_lasso(Y ~ Z, ~ X_noise, data = dat)
)
fallback_summary(fits)
#>           fit adjustment n_selected selected fallback_reason
#> 1 informative        lin          1    X_sig            <NA>
#> 2  noise_only        lin          1  X_noise            <NA>
fallback_summary(fits, only_fallbacks = TRUE)
#> [1] fit             adjustment      n_selected      selected       
#> [5] fallback_reason
#> <0 rows> (or 0-length row.names)