Drift report as a Shiny app

The report of drfthtml() as a Shiny app, for working through a drift result rather than handing it on. drftapp(x) opens on a cpsdrft() result; drftapp() with no result opens on two file inputs, the reference and the current CSV, and compares them when both are uploaded. It needs the shiny package, which is suggested, not required.

Usage

drftapp(
  x = NULL,
  ref = NULL,
  cur = NULL,
  hist = NULL,
  warn = NULL,
  fail = NULL
)

Arguments

x

A cpsdrft() result, or NULL (the default) to upload two CSV files in the app.

ref, cur

Optional: the two data frames x was computed from, for the rerun panel.

hist

Optional: a directory of earlier drift JSON files, for the history tab.

warn, fail

The PSI bands the page opens with; NULL (the default) takes the object's bands.

Value

A Shiny app object (class shiny.appobj).

Details

The app has the column table (verdicts under the bands), the PSI per column, the fingerprint, and for the chosen column its summary, the distributions, the ECDFs, the quantile shift, the first digits and a sentence on the verdict, as drfthtml() has, and a history heatmap with hist; the warn and fail bands and alpha are sliders, with a reset to the object's bands. Beyond the page it has: “what would clear this”, the PSI of the column with its k largest bins left out and the rest renormalized; with ref and cur (or uploaded CSV files), a rerun of cpsdrft() on chosen columns with another number of bins; with hist, the PSI of each column over the runs in that directory (rmbl drift --json outputs or JSON exports of the page); a manifest tab when x$manifest holds an mnfdiff(); and downloads of the HTML report, Markdown and JSON. Uploaded files are read with the package's own CSV reader (the one dload() uses). rmbl drift PATH --app opens the app on the first capsule whose source moved.

The function builds the app and returns it; printing it, or shiny::runApp(), runs it.

Errors: x neither NULL nor from cpsdrft(); shiny not installed; only one of ref and cur, or either not a data frame; hist not a directory or holding no drift JSON; warn above fail.

See also

drfthtml() for a file to hand on.

Examples

set.seed(5)
ref <- data.frame(value = stats::rnorm(400),
                  grade = sample(c("a", "b", "c"), 400, TRUE))
cur <- data.frame(value = stats::rnorm(400, 0.4),
                  grade = sample(c("a", "b", "c"), 400, TRUE))
d <- cpsdrft(ref, cur)
if (FALSE) { # interactive()
if (requireNamespace("shiny", quietly = TRUE)) {
  # drftapp() alone opens on two CSV files to upload
  print(drftapp(d, ref = ref, cur = cur))
}
}