Classifies each line of output as blank, boilerplate, an echo
of the inputs, an optimiser-settings line, content the fitted object
stores, content also delivered through the direct-export channel, or
unaccounted. The last category is the point: it lists what
would be lost if the printed output were ever discarded, and the
package's tests require it to be empty for a corpus of fits.
Arguments
- x
a
tda_result(fromtda_run) or any fitted object carrying one as$run.
Value
a data frame with columns line and class, one
row per output line, with a summary attribute counting each
class.
Examples
d <- data.frame(x = 1:8, y = (1:8) * 2 + rnorm(8))
f <- tda_lsreg(y ~ x, d)
a <- tda_output_audit(f)
attr(a, "summary")
#> cls
#> blank boilerplate echo settings stored
#> 15 9 7 3 12
subset(a, class == "UNACCOUNTED")
#> [1] line class
#> <0 rows> (or 0-length row.names)