Package 'purrrlyr'

Apply a function to each row of a data frame

Description

by_row() and invoke_rows() apply ..f to each row of .d. If ..f's output is not a data frame nor an atomic vector, a list-column is created. In all cases, by_row() and invoke_rows() create a data frame in tidy format.

Usage

by_row(
  .d,
  ..f,
  ...,
  .collate = c("list", "rows", "cols"),
  .to = ".out",
  .labels = TRUE
)

invoke_rows(
  .f,
  .d,
  ...,
  .collate = c("list", "rows", "cols"),
  .to = ".out",
  .labels = TRUE
)

Arguments

.d	A data frame.
...	Further arguments passed to ..f.
.collate	If "list", the results are returned as a list- column. Alternatively, if the results are data frames or atomic vectors, you can collate on "cols" or on "rows". Column collation require vector of equal length or data frames with same number of rows.
.to	Name of output column.
.labels	If TRUE, the returned data frame is prepended with the labels of the slices (the columns in .d used to define the slices). They are recycled to match the output size in each slice if necessary.
.f, ..f	A function to apply to each row. If ..f does not return a data frame or an atomic vector, a list-column is created under the name .out. If it returns a data frame, it should have the same number of rows within groups and the same number of columns between groups.

Details

By default, the whole row is appended to the result to serve as identifier (set .labels to FALSE to prevent this). In addition, if ..f returns a multi-rows data frame or a non-scalar atomic vector, a .row column is appended to identify the row number in the original data frame.

invoke_rows() is intended to provide a version of pmap() for data frames. Its default collation method is "cols", which makes it equivalent to mdply() from the plyr package. Note that invoke_rows() follows the signature pattern of the invoke family of functions and takes .f as its first argument.

The distinction between by_row() and invoke_rows() is that the former passes a data frame to ..f while the latter maps the columns to its function call. This is essentially like using invoke() with each row. Another way to view this is that invoke_rows() is equivalent to using by_row() with a function lifted to accept dots (see lift()).

Value

A data frame.

See Also

by_slice()

Examples

# ..f should be able to work with a list or a data frame. As it
# happens, sum() handles data frame so the following works:
mtcars %>% by_row(sum)

# Other functions such as mean() may need to be adjusted with one
# of the lift_xy() helpers:
mtcars %>% by_row(purrr::lift_vl(mean))

# To run a function with invoke_rows(), make sure it is variadic (that
# it accepts dots) or that .f's signature is compatible with the
# column names
mtcars %>% invoke_rows(.f = sum)
mtcars %>% invoke_rows(.f = purrr::lift_vd(mean))

# invoke_rows() with cols collation is equivalent to plyr::mdply()
p <- expand.grid(mean = 1:5, sd = seq(0, 1, length = 10))
p %>% invoke_rows(.f = rnorm, n = 5, .collate = "cols")
## Not run: 
p %>% plyr::mdply(rnorm, n = 5) %>% dplyr::tbl_df()

## End(Not run)

# To integrate the result as part of the data frame, use rows or
# cols collation:
mtcars[1:2] %>% by_row(function(x) 1:5)
mtcars[1:2] %>% by_row(function(x) 1:5, .collate = "rows")
mtcars[1:2] %>% by_row(function(x) 1:5, .collate = "cols")

---

Apply a function to slices of a data frame

Description

by_slice() applies ..f on each group of a data frame. Groups should be set with slice_rows() or dplyr::group_by().

Usage

by_slice(
  .d,
  ..f,
  ...,
  .collate = c("list", "rows", "cols"),
  .to = ".out",
  .labels = TRUE
)

Arguments

.d	A sliced data frame.
..f	A function to apply to each slice. If ..f does not return a data frame or an atomic vector, a list-column is created under the name .out. If it returns a data frame, it should have the same number of rows within groups and the same number of columns between groups.
...	Further arguments passed to ..f.
.collate	If "list", the results are returned as a list- column. Alternatively, if the results are data frames or atomic vectors, you can collate on "cols" or on "rows". Column collation require vector of equal length or data frames with same number of rows.
.to	Name of output column.
.labels	If TRUE, the returned data frame is prepended with the labels of the slices (the columns in .d used to define the slices). They are recycled to match the output size in each slice if necessary.

Details

by_slice() provides equivalent functionality to dplyr's dplyr::do() function. In combination with map(), by_slice() is equivalent to dplyr::summarise_each() and dplyr::mutate_each(). The distinction between mutating and summarising operations is not as important as in dplyr because we do not act on the columns separately. The only constraint is that the mapped function must return the same number of rows for each variable mapped on.

Value

A data frame.

See Also

by_row(), slice_rows(), dmap()

Examples

# Here we fit a regression model inside each slice defined by the
# unique values of the column "cyl". The fitted models are returned
# in a list-column.
mtcars %>%
  slice_rows("cyl") %>%
  by_slice(purrr::partial(lm, mpg ~ disp))

# by_slice() is especially useful in combination with map().

# To modify the contents of a data frame, use rows collation. Note
# that unlike dplyr, Mutating and summarising operations can be
# used indistinctly.

# Mutating operation:
df <- mtcars %>% slice_rows(c("cyl", "am"))
df %>% by_slice(dmap, ~ .x / sum(.x), .collate = "rows")

# Summarising operation:
df %>% by_slice(dmap, mean, .collate = "rows")

# Note that mapping columns within slices is best handled by dmap():
df %>% dmap(~ .x / sum(.x))
df %>% dmap(mean)

# If you don't need the slicing variables as identifiers, switch
# .labels to FALSE:
mtcars %>%
  slice_rows("cyl") %>%
  by_slice(purrr::partial(lm, mpg ~ disp), .labels = FALSE) %>%
  purrr::flatten() %>%
  purrr::map(coef)

---

Map over the columns of a data frame

Description

dmap() is just like purrr::map() but always returns a data frame. In addition, it handles grouped or sliced data frames.

Usage

dmap(.d, .f, ...)

dmap_at(.d, .at, .f, ...)

dmap_if(.d, .p, .f, ...)

Arguments

.d	A data frame.
.f	A function, formula, or vector (not necessarily atomic). If a function, it is used as is. If a formula, e.g. ~ .x + 2, it is converted to a function. There are three ways to refer to the arguments: For a single argument function, use . For a two argument function, use .x and .y For more arguments, use ..1, ..2, ..3 etc This syntax allows you to create very compact anonymous functions. If character vector, numeric vector, or list, it is converted to an extractor function. Character vectors index by name and numeric vectors index by position; use a list to index by position and name at different levels. If a component is not present, the value of .default will be returned.
...	Additional arguments passed on to the mapped function.
.at	A character vector of names, positive numeric vector of positions to include, or a negative numeric vector of positions to exlude. Only those elements corresponding to .at will be modified. If the tidyselect package is installed, you can use vars() and the tidyselect helpers to select elements.
.p	A single predicate function, a formula describing such a predicate function, or a logical vector of the same length as .x. Alternatively, if the elements of .x are themselves lists of objects, a string indicating the name of a logical element in the inner lists. Only those elements where .p evaluates to TRUE will be modified.

Details

dmap_at() and dmap_if() recycle length 1 vectors to the group sizes.

Examples

# dmap() always returns a data frame:
dmap(mtcars, summary)

# dmap() also supports sliced data frames:
sliced_df <- mtcars[1:5] %>% slice_rows("cyl")
sliced_df %>% dmap(mean)
sliced_df %>% dmap(~ .x / max(.x))

# This is equivalent to the combination of by_slice() and dmap()
# with 'rows' collation of results:
sliced_df %>% by_slice(dmap, mean, .collate = "rows")

---

Slice a data frame into groups of rows

Description

slice_rows() is equivalent to dplyr's dplyr::group_by() command but it takes a vector of column names or positions instead of capturing column names with special evaluation. unslice() removes the slicing attributes.

Usage

slice_rows(.d, .cols = NULL)

unslice(.d)

Arguments

.d	A data frame to slice or unslice.
.cols	A character vector of column names or a numeric vector of column positions. If NULL, the slicing attributes are removed.

Value

A sliced or unsliced data frame.

See Also

by_slice() and dplyr::group_by()

---