Title: A Tidy Interface for Simulating Multivariate Data
Version: 0.2.2
Description: Provides pipe-friendly (%>%) wrapper functions for MASS::mvrnorm() to create simulated multivariate data sets with groups of variables with different degrees of variance, covariance, and effect size.
License: MIT + file LICENSE
Encoding: UTF-8
Imports: dplyr, tibble, MASS, purrr, rlang, assertthat
RoxygenNote: 7.2.3
URL: https://github.com/Aariq/holodeck
BugReports: https://github.com/Aariq/holodeck/issues
Suggests: testthat, covr, knitr, rmarkdown, mice, ggplot2
VignetteBuilder: knitr
NeedsCompilation: no
Packaged: 2023-08-25 21:43:57 UTC; ericscott
Author: Eric Scott ORCID iD [aut, cre]
Maintainer: Eric Scott <scottericr@gmail.com>
Repository: CRAN
Date/Publication: 2023-08-25 22:00:06 UTC

holodeck: A Tidy Interface for Simulating Multivariate Data

Description

Provides pipe-friendly (

Author(s)

Maintainer: Eric Scott scottericr@gmail.com (ORCID)

See Also

Useful links:

Definition operator

Description

Internally, this package uses the definition operator, :=, to make assignments that require computing on the LHS.

Arguments

x

An object to test.

lhs, rhs

Expressions for the LHS and RHS of the definition.

Pipe friendly wrapper to 'diag(x) <- value'

Description

Pipe friendly wrapper to 'diag(x) <- value'

Usage

set_diag(x, value)

Arguments

x

a matrix

value

either a single value or a vector of length equal to the diagonal of 'x'.

Value

a matrix

Examples

library(dplyr)
matrix(0,3,3) %>%
set_diag(1)

Simulate categorical data

Description

This is a simple wrapper that creates a tibble of length 'n_obs' with a single column 'groups'. It will warn if there are fewer than three replicates per group.

Usage

sim_cat(.data = NULL, n_obs = NULL, n_groups, name = "group")

Arguments

.data

An optional dataframe. If a dataframe is supplied, simulated categorical data will be added to the dataframe. Either '.data' or 'n_obs' must be supplied.

n_obs

Total number of observations/rows to simulate if '.data' is not supplied.

n_groups

How many groups or treatments to simulate.

name

The column name for the grouping variable. Defaults to "group".

Details

To-do:

- Make this optionally create multiple categorical variables as being nested or crossed or random

Value

a tibble

See Also

sim_covar, sim_discr

Other multivariate normal functions: sim_covar(), sim_discr()

Examples

df <- sim_cat(n_obs = 30, n_groups = 3)

Simulate co-varying variables

Description

Adds a group of variables (columns) with a given variance and covariance to a data frame or tibble

Usage

sim_covar(.data = NULL, n_obs = NULL, n_vars, var, cov, name = NA, seed = NA)

Arguments

.data

An optional dataframe. If a dataframe is supplied, simulated categorical data will be added to the dataframe. Either '.data' or 'n_obs' must be supplied.

n_obs

Total number of observations/rows to simulate if '.data' is not supplied.

n_vars

Number of variables to simulate.

var

Variance used to construct variance-covariance matrix.

cov

Covariance used to construct variance-covariance matrix.

name

An optional name to be appended to the column names in the output.

seed

An optional seed for random number generation. If 'NA' (default) a random seed will be used.

Value

a tibble

See Also

sim_cat, sim_discr

Other multivariate normal functions: sim_cat(), sim_discr()

Examples

library(dplyr)
sim_cat(n_obs = 30, n_groups = 3) %>%
sim_covar(n_vars = 5, var = 1, cov = 0.5, name = "correlated")

Simulate co-varying variables with different means by group

Description

To-do: make this work with 'dplyr::group_by()' instead of 'group ='

Usage

sim_discr(.data, n_vars, var, cov, group_means, name = NA, seed = NA)

Arguments

.data

A dataframe containing a grouping variable column.

n_vars

Number of variables to simulate.

var

Variance used to construct variance-covariance matrix.

cov

Covariance used to construct variance-covariance matrix.

group_means

A vector of the same length as the number of grouping variables.

name

An optional name to be appended to the column names in the output.

seed

An optional seed for random number generation. If 'NA' (default) a random seed will be used.

Value

a tibble

See Also

sim_cat, sim_covar

Other multivariate normal functions: sim_cat(), sim_covar()

Examples

library(dplyr)
sim_cat(n_obs = 30, n_groups = 3) %>%
group_by(group) %>%
sim_discr(n_vars = 5, var = 1, cov = 0.5, group_means = c(-1, 0, 1), name = "descr")

Simulate missing values

Description

Takes a data frame and randomly replaces a user-supplied proportion of values with 'NA'.

Usage

sim_missing(.data, prop, seed = NA)

Arguments

.data

A dataframe.

prop

Proportion of values to be set to 'NA'.

seed

An optional seed for random number generation. If 'NA' (default) a random seed will be used.

Value

a dataframe with NAs

Examples

library(dplyr)
df <- sim_cat(n_obs = 10, n_groups = 2) %>%
sim_covar(n_vars = 10, var = 1, cov = 0.5) %>%
sim_missing(0.05)