Package 'validatesuggest'

Title: Generate Suggestions for Validation Rules
Description: Generate suggestions for validation rules from a reference data set, which can be used as a starting point for domain specific rules to be checked with package 'validate'.
Authors: Edwin de Jonge [aut, cre] , Olav ten Bosch [aut]
Maintainer: Edwin de Jonge <[email protected]>
License: MIT + file LICENSE
Version: 0.3.2
Built: 2024-07-03 02:15:03 UTC
Source: https://github.com/data-cleaning/validatesuggest

Help Index

Car owners data set (fictitious).

Description

A constructed data set useful for detecting conditinal dependencies.

Usage

car_owner

Format

A data frame with 200 rows and 4 variables. Each row is a person with:

age

age of person

driver_license

has a driver license, only persons older then 17 can have a license in this data set

income

monthly income

owns_car

only persons with a drivers license , and a monthly income > 1500 can own a car

car_color

NA when there is no car

Examples

data("car_owner")

rules <- suggest_cond_rule(car_owner)
rules$rules

Suggest rules

Description

Suggests rules using the various suggestion checks. Use the more specific suggest functions for more control.

Usage

suggest_rules(
  d,
  vars = names(d),
  domain_check = TRUE,
  range_check = TRUE,
  pos_check = TRUE,
  type_check = TRUE,
  na_check = TRUE,
  unique_check = TRUE,
  ratio_check = TRUE,
  conditional_rule = TRUE
)

suggest_all(
  d,
  vars = names(d),
  domain_check = TRUE,
  range_check = TRUE,
  pos_check = TRUE,
  type_check = TRUE,
  na_check = TRUE,
  unique_check = TRUE,
  ratio_check = TRUE,
  conditional_rule = TRUE
)

write_all_suggestions(
  d,
  vars = names(d),
  file = stdout(),
  domain_check = TRUE,
  range_check = TRUE,
  type_check = TRUE,
  pos_check = TRUE,
  na_check = TRUE,
  unique_check = TRUE,
  ratio_check = TRUE,
  conditional_rule = TRUE
)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
domain_check

if TRUE include domain_check
range_check

if TRUE include range_check
pos_check

if TRUE include pos_check
type_check

if TRUE include type_check
na_check

if TRUE include na_check
unique_check

if TRUE include unique_check
ratio_check

if TRUE include ratio_check
conditional_rule

if TRUE include cond_rule
file

file to which the checks will be written to.

Value

returns validate::validator() object with the suggested rules. write_all_suggestions write the rules to file and returns invisibly a named list of ranges for each variable.

task2 dataset

Description

Fictuous test data set from European (ESSnet) project on validation 2017.

Usage

task2

Format

ID

ID

Age

Age of person

Married

Marital status

Employed

Employed or not

Working_hours

Working hours

References

European (ESSnet) project on validation 2017

Suggest a conditional rule

Description

Suggest a conditional rule based on a association rule. This functions derives conditional rules based on the non-existance of combinations of categories in pairs of variables. For each numerical variable a logical variable is derived that tests for positivity. It generates IF THEN rules based on two variables.

Usage

write_cond_rule(d, vars = names(d), file = stdout())

suggest_cond_rule(d, vars = names(d))

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
file

file to which the checks will be written to.

Value

suggest_cond_rule returns validate::validator() object with the suggested rules. write_cond_rule returns invisibly a named list of ranges for each variable.

Examples

data(retailers, package="validate")

# will generate check for all columns in retailers that are
# complete.
suggest_na_check(retailers)
data("car_owner")

rules <- suggest_cond_rule(car_owner)
rules$rules

Suggest a range check

Description

Suggest a range check

Usage

write_domain_check(d, vars = names(d), only_positive = TRUE, file = stdout())

suggest_domain_check(d, vars = names(d), only_positive = TRUE)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
only_positive

if TRUE only numerical values for positive values are included
file

file to which the checks will be written to.

Value

suggest_domain_check returns validate::validator() object with the suggested rules. write_domain_check returns invisibly a named list of checks for each variable.

Examples

data(SBS2000, package="validate")

suggest_range_check(SBS2000)

# checks the ranges of each variable
suggest_range_check(SBS2000[-1], min=TRUE, max=TRUE)

# checks the ranges of each variable
suggest_range_check(SBS2000, vars=c("turnover", "other.rev"), min=FALSE, max=TRUE)

Suggest a check for completeness.

Description

Suggest a check for completeness.

Usage

write_na_check(d, vars = names(d), file = stdout())

suggest_na_check(d, vars = names(d))

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
file

file to which the checks will be written to.

Value

suggest_na_check returns validate::validator() object with the suggested rules. write_na_check write the rules to file and returns invisibly a named list of ranges for each variable.

Examples

data(retailers, package="validate")

# will generate check for all columns in retailers that are
# complete.
suggest_na_check(retailers)

Suggest a range check

Description

Suggest a range check

Usage

write_pos_check(d, vars = names(d), only_positive = TRUE, file = stdout())

suggest_pos_check(d, vars = names(d), only_positive = TRUE)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
only_positive

if TRUE only numerical values for positive values are included
file

file to which the checks will be written to.

Value

suggest_pos_check returns validate::validator() object with the suggested rules. write_pos_check write the rules to file and returns invisibly a named list of checks for each variable.

Examples

data(SBS2000, package="validate")

suggest_range_check(SBS2000)

# checks the ranges of each variable
suggest_range_check(SBS2000[-1], min=TRUE, max=TRUE)

# checks the ranges of each variable
suggest_range_check(SBS2000, vars=c("turnover", "other.rev"), min=FALSE, max=TRUE)

Suggest a range check

Description

Suggest a range check

Usage

write_range_check(d, vars = names(d), min = TRUE, max = FALSE, file = stdout())

suggest_range_check(d, vars = names(d), min = TRUE, max = FALSE)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
min

TRUE or FALSE, should the minimum value be checked?
max

TRUE or FALSE, should the maximum value be checked?
file

file to which the checks will be written to.

Value

suggest_range_check returns validate::validator() object with the suggested rules. write_range_check write the rules to file and returns invisibly a named list of ranges for each variable.

Examples

data(SBS2000, package="validate")

suggest_range_check(SBS2000)

# checks the ranges of each variable
suggest_range_check(SBS2000[-1], min=TRUE, max=TRUE)

# checks the ranges of each variable
suggest_range_check(SBS2000, vars=c("turnover", "other.rev"), min=FALSE, max=TRUE)

Suggest ratio checks

Description

Suggest ratio checks

Usage

write_ratio_check(
  d,
  vars = names(d),
  file = stdout(),
  lin_cor = 0.95,
  digits = 2
)

suggest_ratio_check(d, vars = names(d), lin_cor = 0.95, digits = 2)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
file

file to which the checks will be written to.
lin_cor

threshold for abs correlation to be included (details)
digits

number of digits for rounding

Value

suggest_ratio_check returns validate::validator() object with the suggested rules. write_ratio_check write the rules to file and returns invisibly a named list of check for each variable.

Examples

data(SBS2000, package="validate")

# generates upper and lower checks for the
# ratio of two variables if their correlation is
# bigger then `lin_cor`
suggest_ratio_check(SBS2000, lin_cor=0.98)

suggest type check

Description

suggest type check

Usage

write_type_check(d, vars = names(d), file = stdout())

suggest_type_check(d, vars = names(d))

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
file

file to which the checks will be written to.

Value

suggest_type_check returns validate::validator() object with the suggested rules. write_type_check write the rules to file and returns invisibly a named list of types for each variable.

Suggest range checks

Description

Suggest range checks

Usage

write_unique_check(d, vars = names(d), file = stdout(), fraction = 0.95)

suggest_unique_check(d, vars = names(d), fraction = 0.95)

Arguments

d

data.frame, used to generate the checks
vars

character optionally the subset of variables to be used.
file

file to which the checks will be written to.
fraction

if values in a column > fraction unique, the check will be generated.

Value

suggest_unique_check returns validate::validator() object with the suggested rules. write_unique_check write the rules to file and returns invisibly a named list of checks for each variable.