Skip to contents

The score model each arm is generated from

Source: R/pca.R
pca_scores.Rd

One row per arm and retained component, giving the mean score that arm is centred on and the residual standard deviation it is scattered by. A generated subject's scores are their arm's mean plus a draw whose spread is sd, so these two columns are the whole of the between-subject variability the synthetic data will have.

Usage

pca_scores(x)

Arguments

x

A dataset from synpmx_pca(), or its trial summary.

Value

A data frame with arm, component, mean and sd. Marked "restricted_not_releasable".

Details

The sd values differ by arm on purpose. An arm sitting on an assay limit is genuinely tighter than one well above it, and giving every arm the pooled spread would smear the low arms upward.

Both dose_term settings report the same shape. Under "factor" the mean is the arm's own; under "log" it is the regression evaluated at that arm's planned total dose, and sd is the shared residual.

See also

synpmx_pca_summarize(), pca_components(), pca_report().

Examples

data <- pmx_simulated_fixture(60)
roles <- pmx_roles(
  id = "ID", time = "TIME", nominal_time = "NTIME", dv = "DV", amt = "AMT",
  evid = "EVID", cmt = "CMT", dvid = "DVID", mdv = "MDV"
)
pca_scores(synpmx_pca_summarize(data, roles))
#>   arm component         mean       sd
#> 1 all       PC1 6.443919e-16 3.741645