Package 'messy'

Title: Create Messy Data from Clean Data Frames
Description: For the purposes of teaching, it is often desirable to show examples of working with messy data and how to clean it. This R package creates messy data from clean, tidy data frames so that students have a clean example to work towards.
Authors: Nicola Rennie [aut, cre, cph] (ORCID: <https://orcid.org/0000-0003-4797-557X>)
Maintainer: Nicola Rennie <[email protected]>
License: CC BY 4.0
Version: 0.1.0.9003
Built: 2026-05-26 07:50:06 UTC
Source: https://github.com/nrennie/messy

Help Index

Add special characters to strings

Description

Add special characters to strings

Usage

add_special_chars(data, cols = NULL, messiness = 0.1)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all columns. Default NULL.
messiness

Percentage of values to change. Must be between 0 and 1. Default 0.1.

Value

a dataframe the same size as the input data.

Examples

add_special_chars(mtcars)

Add whitespaces

Description

Randomly add whitespaces to the end of some values in all or a subset of columns.

Usage

add_whitespace(data, cols = NULL, messiness = 0.1)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all columns. Default NULL.
messiness

Percentage of values to change. Must be between 0 and 1. Default 0.1.

Value

a dataframe the same size as the input data.

Examples

add_whitespace(mtcars)

Change case

Description

Randomly switch between title case and lowercase for character strings

Usage

change_case(data, cols = NULL, messiness = 0.1, case_type = "word")

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all columns. Default NULL.
messiness

Percentage of values to change. Must be between 0 and 1. Default 0.1.
case_type

Whether the case should change based on the "word" or "letter".

Value

a dataframe the same size as the input data.

Examples

change_case(mtcars)

Change separators

Description

Randomly change the separators in character strings through random replacement

Usage

change_separators(
  data,
  cols = NULL,
  messiness = 0.1,
  sep_in = c("-", "_", "  ", " "),
  sep_out = c("-", "_", "  ", " ")
)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all columns. Default NULL.
messiness

Percentage of values to change. Must be between 0 and 1. Default 0.1.
sep_in

A single value, or vector, or list of what is considered a separator in the input data. Default c("-", "_", " ", " ").
sep_out

A single value, or vector, or list of what the separators may be randomly with. Default c("-", "_", " ", " ").

Value

a dataframe the same size as the input data.

Examples

change_separators(mtcars)

Duplicate columns and insert them into the dataframe at random

Description

Duplicate columns and insert them into the dataframe at random

Usage

duplicate_columns(data, messiness = 0.1, random = TRUE, name_sep = "")

Arguments

data

input dataframe
messiness

Probability that each column is duplicated. Must be between 0 and 1. Default 0.1.
random

Whether duplicated column names should be randomly selected from other column names, or maintain the original. Default TRUE.
name_sep

Separator to use for adding numbers to end of names. Default "".

Value

A dataframe with duplicated rows inserted

Author(s)

Jordi Rosell

Examples

duplicate_columns(mtcars, messiness = 0.1)

Duplicate rows and insert them into the dataframe in order or at random. May result in numbers being added to the end of row names.

Description

Duplicate rows and insert them into the dataframe in order or at random. May result in numbers being added to the end of row names.

Usage

duplicate_rows(data, messiness = 0.1, shuffle = FALSE)

Arguments

data

input dataframe
messiness

Percentage of rows to duplicate. Must be between 0 and 1. Default 0.1.
shuffle

Insert duplicated data underneath original data or insert randomly

Value

A dataframe with duplicated rows inserted

Author(s)

Philip Leftwich, Barry Rowlingson

Examples

duplicate_rows(mtcars, messiness = 0.1)

Make missing

Description

Randomly make values missing in all data columns, or a subset of columns

Usage

make_missing(data, cols = NULL, messiness = 0.1, missing = NA)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all columns. Default NULL.
messiness

Percentage of values to change. Must be between 0 and 1. Default 0.1.
missing

A single value, vector, or list of what the missing values will be replaced with. If length is greater than 1, values will be replaced randomly. Default NA.

Value

a dataframe the same size as the input data.

Examples

make_missing(mtcars)

Messy

Description

Make a data frame messier.

Usage

messy(
  data,
  messiness = 0.1,
  missing = NA,
  case_type = "word",
  sep_in = c("-", "_", "  ", " "),
  sep_out = c("-", "_", "  ", " ")
)

Arguments

data

input dataframe
messiness

Percentage of values to change per function. Must be between 0 and 1. Default 0.1.
missing

A single value, vector, or list of what the missing values will be replaced with. If length is greater than 1, values will be replaced randomly. Default NA.
case_type

Whether the case should change based on the "word" or "letter".
sep_in

A single value, or vector, or list of what is considered a separator in the input data. Default c("-", "_", " ", " ").
sep_out

A single value, or vector, or list of what the separators may be randomly with. Default c("-", "_", " ", " ").

Value

a dataframe the same size as the input data.

Examples

messy(mtcars)

Make column names messy

Description

Adds special characters and randomly capitalises characters in the column names of a data frame.

Usage

messy_colnames(data, messiness = 0.2)

Arguments

data

data.frame to alter column names
messiness

Percentage of values to change per function. Must be between 0 and 1. Default 0.1.

Value

data.frame with messy column names

Author(s)

Athanasia Monika Mowinckel

Examples

messy_colnames(mtcars)

Make date(time) formats inconsistent

Description

Takes any date(times) column and transforms it into a character column, sampling from any number of random of valid character representations.

Usage

messy_datetime_formats(
  data,
  cols = NULL,
  formats = c("%Y/%m/%d %H:%M:%S", "%d/%m/%Y %H:%M:%S")
)

messy_date_formats(
  data,
  cols = NULL,
  formats = c("%Y/%m/%d", "%d/%m/%Y")
)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all POSIXt columns (for messy_datetime_formats()) or Date columns (for messy_date_formats()).
formats

A vector of any number of valid strptime() formats. Multiple formats will be sampled at random.

Value

a dataframe the same size as the input data.

Author(s)

Jack Davison

See Also

Other Messy date(time) functions: messy_datetime_tzones(), split_datetimes()

Examples

data <- data.frame(dates = rep(Sys.Date(), 10))
messy_date_formats(data)

Change the timezone of datetime columns

Description

Takes any number of datetime columns and changes their timezones either totally at random, or from a user-provided list of timezones.

Usage

messy_datetime_tzones(data, cols = NULL, tzones = OlsonNames(), force = FALSE)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all POSIXt columns.
tzones

Valid time zones to sample from. By default samples from all OlsonNames(), but can be set to options more relevant to the data.
force

By default (force = FALSE) the datetimes will have their actual hour/minute values changed along with the timezones. If force = TRUE, which requires lubridate, the datetime values will remain the same and only the timezone will differ.

Value

a dataframe the same size as the input data.

Author(s)

Jack Davison

See Also

Other Messy date(time) functions: messy_datetime_formats(), split_datetimes()

Examples

data <- data.frame(dates = rep(Sys.time(), 10))

data$dates
attr(data$dates, "tzone")

messy <- messy_datetime_tzones(data, tzones = "Poland")
messy$dates
attr(messy$dates, "tzone")

Splits date(time) column(s) into multiple columns

Description

These functions can split the "date" and "time" components of POSIXt columns and the "hour", "month", and "day" components of Date columns into multiple columns.

Usage

split_datetimes(data, cols = NULL, class = c("character", "date"))

split_dates(data, cols = NULL)

Arguments

data

input dataframe
cols

set of columns to apply transformation to. If NULL will apply to all POSIXt columns (for split_datetimes()) or Date columns (for split_dates()).
class

For split_datetimes(). The desired output of the separate "date" and "time" columns. "character" leaves the columns as character vectors. "date" will reformat the date as a "Date" and the time as a "POSIXct" object, with a dummy date appended to it. In split_dates(), all returned columns are integers.

Value

a dataframe

Author(s)

Jack Davison

See Also

Other Messy date(time) functions: messy_datetime_formats(), messy_datetime_tzones()

Examples

# split datetimes
data <- data.frame(today = Sys.time())
split_datetimes(data)
# split dates
data <- data.frame(today = Sys.Date())
data
split_dates(data)

Splits a dataframe into two, such that it could be reassembled with a mutating join

Description

This function takes an arbitrary number of 'joining' columns and any number of additional column names and splits a dataframe in two such that a user could then re-join using merge() or dplyr::left_join(). The user may find it appropriate to go on and apply messy() to each new dataframe independently to impede rejoining.

Usage

unjoin(data, by, cols, distinct = "none", names = c("left", "right"))

Arguments

data

input dataframe
by

a vector of column names which will be present in both outputs, to rejoin the dataframes
cols

specific columns to be present in the 'right' dataframe. implicitly, all other columns not in 'cols' will be present in the 'left' dataframe.
distinct

Apply dplyr::distinct() to "both" dataframes, the "left" or "right" dataframes, or "none" of the dataframes. This may be useful if one table is a 'lookup' or metadata table that has its values repeated many times in data.
names

The names of the output list. If NULL the list will be unnamed.

Details

Real data is often found across multiple datasets. For example, in environmental monitoring, measurements at a monitoring station may need to be bound with metadata about the station such as geographic coordinates, or even meteorological data from an external source, to produce desired outputs. In clinical research it may be necessary to combine the results of a clinical trial with relevant patient information, such as weight or sex. This function undoes existing joins to present learners with an authentic problem to solve; joining two independent datasets to achieve some goal.

Value

A list of two dataframes

Author(s)

Jack Davison

See Also

Other data deconstructors: unrbind()

Examples

dummy <-
  dplyr::tibble(
    patient_id = c(1, 1, 1, 2, 2, 2, 3, 3, 3),
    test = c(1, 2, 3, 1, 2, 3, 1, 2, 3),
    result = c("++", "+", "-", "--", "+", "-", "+", "++", "-"),
    sex = c("M", "M", "M", "M", "M", "M", "F", "F", "F"),
    age = c(50, 50, 50, 25, 25, 25, 30, 30, 30)
  )

unjoin(
  dummy,
  by = "patient_id",
  cols = c("sex", "age"),
  distinct = "right",
  names = c("tests", "patient_info")
)

Splits a dataframe row-wise or col-wise into any arbitrary number of dataframes

Description

This function splits a dataframe into any number of dataframes such that they can be rejoined by using rbind()/dplyr::bind_rows() for unrbind() or cbind()/dplyr::bind_cols() for uncbind(). The user may find it appropriate to go on and apply messy() to each new dataframe independently to impede rejoining.

Usage

unrbind(data, sizes = NULL, probs = NULL, names = NULL, shuffle = TRUE)

uncbind(data, sizes = NULL, probs = NULL, names = NULL, shuffle = TRUE)

Arguments

data

input dataframe
sizes

A vector of numeric inputs summing to nrow(data) for unrbind() or ncol(data) for uncbind(); the number of rows of each resulting dataframe. See probs for an alternative approach. If neither are provided, the dataframe will be split roughly in half.
probs

A vector of numeric inputs summing to 1; the proportion of rows/columns in each resulting dataframe. An alternative to sizes.
names

The names of the output list. If NULL the list will be unnamed.
shuffle

Shuffle rows in unrbind() or columns in uncbind()? Defaults to TRUE.

Details

Real data can often be found in disparate files. For example, data reports may come in monthly and require row-binding together to obtain a complete annual time series. Scientific results may arrive from different laboratories and require binding together for further analysis and comparisons. This function may simulate a single dataframe having come from different sources and requiring binding back together. Base R's split() offers an alternative to unrbind(), but requires a pre-existing factor column to split by and cannot as easily create random splits in the data.

Value

A list of dataframes

Author(s)

Jack Davison

See Also

Other data deconstructors: unjoin()

Examples

unrbind(dplyr::tibble(mtcars), probs = c(0.5, 0.3, 0.2))

uncbind(dplyr::tibble(mtcars), probs = c(0.5, 0.3, 0.2))