str2str (version 1.0.0)

lv2d: List of (atomic) vectors to Data-Frame

Description

lv2d converts a list of (atomic) vectors to a data.frame. This function is similar to as.data.frame.list, but allows for more flexibility in how the data.frame will be structured (e.g., rowwise), while simplifying the dimension naming process.

Usage

lv2d(
  lv,
  along,
  fill = FALSE,
  risky = FALSE,
  stringsAsFactors = FALSE,
  check = TRUE
)

Value

data.frame with the elements of `lv` either as rows or columns and dimnames determined along the names of `lv` and `lv`[[1]].

Arguments

lv

list of (atomic) vectors.

along

numeric vector of length 1 specifying either 1 for binding along rows (i.e., each list element is a row) or 2 for binding along columns (i.e., each list element in a column).

fill

logical vector of length 1 specifying whether 1) to allow the vectors in lv to have different lengths, names, or both, 2) to bind by the names of the vectors within lv rather than by their positions (unless no names are present in which case positions are used), and 3) fill in any missing values in the return object with NA.

risky

logical vector of length 1 specifying whether to use list2DF rather than data.frame when along = 2 and fill = TRUE. If either along = 1 or fill = FALSE, this argument is not used.

stringsAsFactors

logical vector of length 1 specifying whether character vectors should be coerced to factors. See default.stringsAsFactors.

check

logical vector of length 1 specifying whether to check the structure of the input arguments. For example, check whether lv is a list of atomic vectors. This argument is available to allow flexibility in whether the user values informative error messages (TRUE) vs. computational efficiency (FALSE).

Details

If fill = FALSE, lv2d uses a combination of do.call and rbind if along = 1 or do.call and cbind if along = 2. rownames and colnames of the returned data.frame are determined by the names of lv and the names of the first vector within lv. If either are NULL, then the positions are used as the dimension names. If fill = FALSE, then an error is returned ff the vectors in lv do not all have the same length. If fill = FALSE, there is no check to ensure the elements within each lv vector have the same names in the same order. The names are taken from the first vector in lv, and it is assumed those names and their order apply to each vector in lv. Essentially, if fill = FALSE, lv binds the vectors by positions and not names.

If fill = TRUE, lv2d uses plyr::rbind.fill if along = 1 or plyr::join_all by the vector names if along = 2. If fill = TRUE, lv2d binds the vectors by by names (and by positions if no names are present). Depending on what the user wants, fill = FALSE or TRUE could be safer. If the user wants an error returned when any vectors within lv have different lengths, then fill = FALSE should be used. If the user wants to bind by names rather than position, then fill = TRUE should be used.

Examples

# 1) `lv` has names; vectors have names
lv <- setNames(object = lapply(X = letters, FUN = setNames, nm = "alphabet"), nm = LETTERS)
lv2d(lv, along = 1)
lv2d(lv, along = 2)
lv2d(lv, along = 2, stringsAsFactors = TRUE)

# 2) `lv` has names; no vector names
lv <- setNames(object = as.list(letters), nm = LETTERS)
lv2d(lv, along = 1)
lv2d(lv, along = 2)

# 3) no `lv` names; vector have names
lv <- lapply(X = letters, FUN = setNames, nm = "alphabet")
lv2d(lv, along = 1)
lv2d(lv, along = 2)

# 4) no `lv` names; no vector names
lv <- as.list.default(letters)
lv2d(lv, along = 1)
lv2d(lv, along = 2)

# we want vectors combined along rows
lv <- lapply(X = unclass(mtcars), FUN = `names<-`, value = row.names(mtcars))
rbind(lv) # not what we want (array list)
rbind.data.frame(lv) # also not what we want (combined along cols)
do.call(what = rbind.data.frame, args = lv) # doesn't have useful dimnames
lv2d(lv, along = 1) # finally what we want

# fill = TRUE
tmp <- lapply(X = unclass(mtcars), FUN = `names<-`, value = row.names(mtcars))
lv <- lapply(X = tmp, FUN = function(v) v[-(sample(x = seq_along(v), size = 9))])
lv2d(lv, along = 1L, fill = TRUE) # NA for missing values in any given row
tmp <- lapply(X = unclass(as.data.frame(t(mtcars))), FUN = `names<-`, value = names(mtcars))
lv <- lapply(X = tmp, FUN = function(v) v[-(sample(x = seq_along(v), size = 3))])
lv2d(lv, along = 2L, fill = TRUE) # NA for missing values in any given column

# actual use case
lv <- lapply(X = sn(1:30), FUN = function(i)
   coef(lm(v2frm(names(attitude)), data = attitude[-i, ])))
lv2d(lv, along = 2) # coefs in a data.frame

# when vectors have named elements in different positions use fill = TRUE
lv <- list("row_1" = c("col_A" = "col_A1", "col_B" = "col_B1", "col_C" = "col_C1"),
"row_2" = c("col_B" = "col_B2", "col_C" = "col_C2", "col_A" = "col_A2"),
"row_3" = c("col_C" = "col_C3", "col_A" = "col_A3", "col_B" = "col_B3"))
lv2d(lv, along = 1, fill = FALSE) # probably not what you want (See details)
lv2d(lv, along = 1, fill = TRUE) # what we want

# when you have a list with only one vector
lv <- list("A" = c("one" = 1, "two" = 2, "three" = 3))
x <- lv2m(lv, along = 1, fill = FALSE)
y <- lv2m(lv, along = 1, fill = TRUE)
identical(x, y)