lplyr: dplyr verbs for lists and other verbs for data frames

Installation

You can install 'lplyr' from GitHub with:

# install.packages("devtools")
devtools::install_github("paulponcet/lplyr")

Verbs for lists and pairlists

The package 'lplyr' extends some dplyr verbs to lists and pairlists:

library(lplyr)

xs <- list(x1 = 1:3, 
           x2 = 2:5, 
           x3 = list("alpha", c("beta", "gamma")))
           
mutate(xs, x4 = 4)
rename(xs, x0 = x1)
select(xs, -x3)
transmute(xs, x5 = 5)

Usual verbs made for standard evaluation work as well:

mutate_(xs, x4 = ~ 4)
rename_(xs, x0 = ~ x1)
select_(xs, ~ (-x3))
transmute_(xs, x5 = ~ 5)

New verbs for data frames

The mutate_which and transmute_which functions are made for adding new variables or modifying existing ones on a subset of the data.

df <- mtcars[1:10,]
mutate_which(df, gear==4, carb = 100)
transmute_which(df, gear==4, carb = 100)

There is also a standard evaluation version of these functions, called mutate_which_ and transmute_which_:

mutate_which_(df, ~ gear==4, carb = ~ 100)
transmute_which_(df, ~ gear==4, carb = ~ 100)

The function pull selects a column in a data frame and transforms it into a vector. This is useful to use it in combination with magrittr's pipe operator and dplyr's verbs.

mtcars[["mpg"]]
mtcars %>% pull(mpg)

# more convenient than (mtcars %>% filter(mpg > 20))[[3L]]
mtcars %>%
 filter(mpg > 20) %>%
 pull(3)

Chunck

The function chunck is identical to the partition function from package 'multidplyr', except that it adds a logical argument ok: if ok=FALSE, no partition is applied, the input dataset is returned unchanged. This is convenient in programming to easily test the benefit of parallelizing or switch from non-parallel to parallel programming.

options(parallelize = FALSE)

mtcars %>% 
  chunck(ok = getOption("parallelize")) %>% 
  mutate(cyl2 = 2 * cyl) %>% 
  filter(vs == 1) %>% 
  summarise(n()) %>% 
  select(-cyl)