Package 'variables'

General Information on the variables Package

Description

The variables package offers a small collection of objects describing conceptual variables and corresponding methods, for example for generating a grid of values for a (yet) unmeasured variable.

The package was written to support the basefun and mlt packages and will be of limited use outside these packages.

Author(s)

This package is authored by Torsten Hothorn <[email protected]>.

References

Torsten Hothorn (2018), Most Likely Transformations: The mlt Package, Journal of Statistical Software, forthcoming. URL: https://cran.r-project.org/package=mlt.docreg

---

Accessor Functions

Description

Access properties of variable objects

Usage

## S3 method for class 'var'
variable.names(object, ...)
desc(object)
unit(object)
support(object)
bounds(object)
is.bounded(object)

Arguments

object	a variable object
...	additional arguments, currently not used

Details

These generics have corresponding methods for factor_var, ordered_var and numeric_var objects as well as for vars collections of those.

---

Check Observations Against Formal Description

Description

Check if observations correspond to their formal descriptions

Usage

check(object, data)

Arguments

object	an object of class var or vars
data	a data.frame

Details

The function returns true of data matches the description in object.

---

Unordered Categorical Variable

Description

Formal description of an unordered categorical variable

Usage

factor_var(name, desc = NULL, levels, ...)

Arguments

name	character, the name of the variable
desc	character, a description of what is measured
levels	character, the levels of the factor
...	ignored

Details

A conceptual description of a (yet) unobserved unordered categorical variable.

Value

An object of class factor\_var inheriting from var with corresponding methods.

Examples

factor_var("eye", "eye color", c("blue", "brown", "green", "grey", "mixed"))

---

Generate Grids of Observations

Description

Make a grid of values

Usage

mkgrid(object, ...)
## S3 method for class 'continuous_var'
mkgrid(object, n = 2, add = TRUE, ...)

Arguments

object	an object of class var or vars
n	number of grid points for a continous variable
add	logical, adds the add argument (in numeric_var) to support if TRUE
...	additional arguments

Details

The function returns a names list of values for each variable.

---

Numeric Variable

Description

Formal description of numeric variable

Usage

numeric_var(name, desc = NULL, unit = NULL, support = c(0, 1), add = c(0, 0), 
            bounds = NULL, ...)

Arguments

name	character, the name of the variable
desc	character, a description of what is measured
unit	character, the measurement unit
support	the support of the measurements, see below
add	add these values to the support before generating a grid via mkgrid
bounds	an interval defining the bounds of a real sample space
...	ignored

Details

A numeric variable can be discrete (support is then the set of all possible values, either integer or double; integers of length 2 are interpreted as all integers inbetween) or continuous (support is a double of length 2 giving the support of the data).

If a continuous variable is bounded, bounds defines the corresponding interval.

Value

An object of class numeric\_var inheriting from var with corresponding methods.

Examples

numeric_var("age", "age of patient", "years", support = 25:75)

  numeric_var("time", "survival time", "days", support = 0:365)

  numeric_var("time", "survival time", "days", support = c(0.0, 365), 
              bounds = c(0, Inf))

---

Ordered Categorical Variable

Description

Formal description of an ordered categorical variable

Usage

ordered_var(name, desc = NULL, levels, sparse = FALSE, ...)

Arguments

name	character, the name of the variable
desc	character, a description of what is measured
levels	character, the ordered levels of the factor
sparse	logical, set-up a sparse model matrix
...	ignored

Details

A conceptual description of a (yet) unobserved ordered categorical variable.

Value

An object of class ordered\_var inheriting from var with corresponding methods.

Examples

ordered_var("temp", "temperature", c("cold", "lukewarm", "warm", "hot"))

---

Multiple Abstract Descriptions

Description

Concatenate or generate multiple variable descriptions

Usage

## S3 method for class 'var'
c(...)
as.vars(object)

Arguments

object	an object
...	a list of variable objects

Details

c() can be used to concatenate multiple variable objects; the corresponding generics also work for the resulting object. as.vars() tries to infer a formal description from data.

Examples

f <- factor_var("x", levels = LETTERS[1:3])
   n <- numeric_var("y")

   fn <- c(f, n)
   variable.names(fn)
   support(fn)
   is.bounded(fn)
   mkgrid(fn, n = 9)

   as.vars(iris)

---