additive_summary — additive

Compute posterior additive summary

additive_summary(
  summaryCall,
  fhatSamples,
  fhat = rowMeans(fhatSamples),
  df = NULL,
  alpha = 0.05,
  fast = TRUE,
  quants = seq(0, 1, by = 0.005),
  grid_size = 100,
  verbose = FALSE,
  return_samples = TRUE,
  meta = NA
)

Arguments

summaryCall: A gam fomula for the additive summary to be computed. Should be in the form of fhat ~ s(x1) + s(x2) + .... See ?mgcv::formula.gam
fhatSamples: N NMC matrix of posterior draws of the function f, where N is the number of observations and NMC is the number of Monte Carlo posterior samples
fhat: A point estimate (posterior mean) for the function f
df: The dataframe from which the summary will be computed. This should include all the inputs of f
alpha: The function will return the alpha/2 and 1-alpha/2 credible intervals for the summary.
fast: If TRUE, the function will compute the summary on a grid, specifically for quantiles of the covariates as specified in quants, of the data rather than for the whole dataset
quants: The quantiles of the covariates in df on which to compute the summary when fast=TRUE
grid_size
verbose: If TRUE, the function will print out the progress of summary computation
return_samples: If TRUE, the function will of the design matrix for the summary
meta: A tag for the dataframe of the summary

Details

This function computes the point estimate and credible intervals for the summary of the function f. At a minimum, the user must specify the form of the summary, a matrix of posterior draws of f, and the dataframe which contains the inputs of f

Author

Spencer Woody