Skip to contents

Extract a tidy estimates frame from a fitted RdistNN object

Source: R/plot-data.R
plot_data_estimates.Rd

Returns one row per target point and coefficient with the point estimate, stored jackknife standard error, and symmetric normal confidence bounds. This is the data layer of the plotting interface: plot builders consume these frames, and users can consume them directly for custom figures.

Usage

plot_data_estimates(object, level = 0.95, ...)

# Default S3 method
plot_data_estimates(object, level = 0.95, ...)

# S3 method for class 'np_result'
plot_data_estimates(object, level = 0.95, ...)

# S3 method for class 'cate_result'
plot_data_estimates(object, level = 0.95, ...)

# S3 method for class 'plqr_result'
plot_data_estimates(object, level = 0.95, quantity = c("moment", "theta"), ...)

# S3 method for class 'gmm_result'
plot_data_estimates(object, level = 0.95, ...)

Arguments

object

A fitted result object (np_result, cate_result, plqr_result, or gmm_result; tdnn_np_result dispatches through np_result).

level

Confidence level for the normal bounds (default 0.95).

...

Passed to methods; not used.

quantity

For plqr_result fits, which quantity to extract: "moment" (default) uses the localized PLQR moments and their jackknife moment_se; "theta" uses the supplied coefficient values (repeated theta_eval) and the direct-sandwich theta_se. While no root solver exists, moments are the target-specific quantity, so they are the default. The choice is recorded in attr(., "quantity").

Value

A data.frame with columns target_id (integer, 1..m), coef (character; "mu"/"tau" for scalar fits, moment or coefficient names for vector fits), estimate, se, ci_lower, ci_upper, and — when the fit stores evaluation points — target coordinates x0_1..x0_p, with the original covariate names recorded in attr(., "x0_names"). Fits without stored standard errors return NA in se and the bound columns. The confidence level is recorded in attr(., "level").

Experimental methods

Methods for plqr_result objects are experimental, as is all PLQR functionality in this release; see plqr_dnn().

Examples

set.seed(1)
n <- 60
x_mat <- matrix(rnorm(n * 2), n, 2)
y_vec <- x_mat[, 1] + rnorm(n, sd = 0.3)
fit <- np_dnn(
  X = x_mat, X0 = rbind(c(0, 0), c(1, 0)), Y = y_vec,
  k = 1L, s = 10L, se = FALSE
)
plot_data_estimates(fit)
#>   target_id coef    estimate se ci_lower ci_upper x0_1 x0_2
#> 1         1   mu 0.002292107 NA       NA       NA    0    0
#> 2         2   mu 0.832118414 NA       NA       NA    1    0