Detect the range for numerical variables

Detect for each numerical variable in a validation rule set, what its maximum and minimum values are. This allows for manual rule set checking: does rule set x overly constrain numerical values?

detect_boundary_num(x, eps = 1e-08, ...)

Arguments

x

validator object, rule set to be checked

eps

detected fixed values will have this precission.

...

currently not used

Value

data.frame with columns "variable", "lowerbound", "upperbound".

Details

This procedure only finds minimum and maximum values, but misses gaps.

References

Statistical Data Cleaning with R (2017), Chapter 8, M. van der Loo, E. de Jonge

Simplifying constraints in data editing (2015). Technical Report 2015|18, Statistics Netherlands, J. Daalmans

See also

detect_fixed_variables

Other feasibility: detect_boundary_cat(), detect_infeasible_rules(), is_contradicted_by(), is_infeasible(), make_feasible()

Examples

rules <- validator(
  x >= 1,
  x + y <= 10,
  y >= 6
)

detect_boundary_num(rules)
#>   variable lowerbound upperbound
#> x        x          1          4
#> y        y          6          9

rules <- validator(
  job %in% c("yes", "no"),
  if (job == "no") income == 0,
  income > 0
)

detect_boundary_cat(rules)
#> $job
#> [1] "yes"
#>