Package 'trimmer'

Adjust Data Table with Candidate Elements for Elimination

Description

Adjusts positions of all candidates for elimination in data.table after removing a candidate (due to the fact, that the positions may shift).

Usage

adjust_candidates(cand, cand_top_idx)

Arguments

cand	data.table with candidates for elimination given by their position indices.
cand_top_idx	numeric position index of candidate to be removed.

Value

data.table candidates after any adjustments to position indices of candidates.

---

Convert Numbered Index to Named Index of List Element

Description

Convert Numbered Index to Named Index of List Element

Usage

convert_idx_to_name(vec, obj)

Arguments

vec	numeric numeric index of list element.
obj	list

Value

character named index of list element.

Examples

d <- list(a = list(b = list(c = 3, d = 5), e = c(2,4)))
num_idx <- c(1,1,2)
convert_idx_to_name(num_idx, d)

---

Fix til at undgå R CMD check notes for "no visible binding for global variable"

Description

Dette script gør det muligt at referere til kolonner i data frames ved hjælp af Non Standard Evaluation (NSE) i databehandlingspakker som data.table og dplyr, uden at dette medfører R CMD check notes angående "no visible binding for global variable". Navnene på de variable, der refereres til ved hjælp af NSE, skal blot angives i en vektor til funktionen globalVariables() nedenfor.

Usage

fix_undefined_global_vars()

Details

Dette er den anbefalede løsning fra CRAN.

---

Compute Results From Function Call with Object as Argument

Description

Compute Results From Function Call with Object as Argument

Usage

get_results_for_object(obj, obj_arg_name, fun, ...,
  tolerate_warnings = TRUE)

Arguments

obj	list R object to be trimmed. _MUST_ inherit from the 'list' class.
obj_arg_name	character what is the name of the parameter, that 'obj' must be set to, when invoking 'fun'. Defaults to NULL, in which case the function assumes, that the 'obj' matches the first parameter of 'fun'.
fun	function function that must return the same results, when invoked with 'obj' both before and after trimming.
...	other (named) arguments for 'fun'.
tolerate_warnings	logical tolerate warnings (=TRUE) Or not (=FALSE) from function call results?

Value

results from function call.

---

Convert Size in Bytes to Print Friendly String

Description

Convert Size in Bytes to Print Friendly String

Usage

pf_obj_size(x, digits = 2)

Arguments

x	numeric object size in digits.
digits	numeric number of digits you want.

Value

character priend friendly string.

Examples

pf_obj_size(10)
pf_obj_size(1010)
pf_obj_size(2e06)

---

Trim an R Object

Description

Trims an R object whilst presuming the results of a given function call, where the R object is given as an argument. One popular example could be trimming an R model object whilst presuming the results of the predict function on a sample of data.

Usage

trim(obj, obj_arg_name = NULL, fun = predict, size_target = 0,
  tolerate_warnings = FALSE, verbose = TRUE, dont_touch = list(),
  ...)

Arguments

obj	list R object to be trimmed. _MUST_ inherit from the 'list' class.
obj_arg_name	character what is the name of the parameter, that 'obj' must be set to, when invoking 'fun'. Defaults to NULL, in which case the function assumes, that the 'obj' matches the first parameter of 'fun'.
fun	function function that must return the same results, when invoked with 'obj' both before and after trimming.
size_target	numeric desired maximum size in _MegaBytes_ of object after trimming has been conducted. When this size is achieved, the trimming stops. Defaults to 0, in which case trimming continues, until no further trimming can be done without breaking results from 'fun'.
tolerate_warnings	logical tolerate warnings (=TRUE) Or not (=FALSE) from function call results?
verbose	logical print messages?
dont_touch	list list with name indices of elements, that must not be removed from object by trimming procedure.
...	other (named) arguments for 'fun'.

Examples

# get training data for predictive model.
trn <- datasets::mtcars

# estimate model.
mdl <- lm(mpg ~ ., data = trn)
trim(obj = mdl, obj_arg_name = "object", fun = predict, newdata = trn)
trim(obj = mdl, obj_arg_name = "object", fun = predict, newdata = trn,
dont_touch = list(c("model"), c("qr","tol")))

---