correctRounding: Correct records under linear restrictions for rounding errors

In deducorrect: Deductive Correction, Deductive Imputation, and Deterministic Correction

Description

This algorithm tries to detect and repair records that violate linear (in)equality constraints by correcting possible rounding errors as described by Scholtus(2008). Typically data is constrainted by Rx=a and Qx ≥ b.

Usage

correctRounding(E, dat, ...)

## S3 method for class 'editset'
correctRounding(E, dat, ...)

## S3 method for class 'editmatrix'
correctRounding(E, dat, fixate = NULL, delta = 2,
  K = 10, round = TRUE, assumeUnimodularity = FALSE, ...)

Arguments

E	editmatrix or editset as generated by the editrules package.
dat	data.frame with the data to be corrected
...	arguments to be passed to other methods.
fixate	character with variable names that should not be changed.
delta	tolerance on checking for rounding error
K	number of trials per record. See details
round	should the solution be rounded, default TRUE
assumeUnimodularity	If FALSE, a test is performed before corrections are computed (expensive).

Details

The algorithm first finds violated constraints |r'_{i}x-a_i| > 0 , and selects edits that may be due to a rounding error 0 < |r'_{i}x-a_i| ≤q δ. The algorithm then makes a correction suggestion where the errors are attributed to randomly selected variables under the lineair equality constraints. It checks if the suggested correction does not violate the inequality matrix Q. If it does, it will try to generate a different solution up till K times.

Value

A deducorrrect object.

References

Scholtus S (2008). Algorithms for correcting some obvious inconsistencies and rounding errors in business survey data. Technical Report 08015, Statistics Netherlands.

See Also

deducorrect-object status

Examples

E <- editmatrix(expression( 
    x1 + x2 == x3,
    x2 == x4,
    x5 + x6  + x7 == x8,
    x3 + x8 == x9,
    x9 - x10 == x11
    )
)

dat <- data.frame( x1=12
                 , x2=4
                 , x3=15
                 , x4=4
                 , x5=3
                 , x6=1
                 , x7=8
                 , x8=11
                 , x9=27
                 , x10=41
                 , x11=-13
                 )

sol <- correctRounding(E, dat)


# example with editset
for ( d in dir("../pkg/R/",full.names=TRUE) ) dmp <- source(d)
E <- editmatrix(expression(
    x + y == z,
    x >= 0,
    y >= 0,
    z >= 0,
    if ( x > 0 ) y > 0
    ))
dat <- data.frame(
    x = 1,
    y = 0,
    z = 1)
# solutions causing new violations of conditional rules are rejected 
sol <- correctRounding(E,dat)

# An example with editset
E <- editset(expression(
    x + y == z,
    x >= 0,
    y > 0,
    y < 2,
    z > 1,
    z < 3,
    A %in% c('a','b'),
    B %in% c('c','d'),
    if ( A == 'a' ) B == 'b',
    if ( B == 'b' ) x < 1
))
dat <- data.frame(
    x = 0,
    y = 1,
    z = 2,
    A = 'a',
    B = 'b'
)

correctRounding(E,dat)    

deducorrect documentation built on May 2, 2019, 3:47 p.m.