In elbersb/tidylog: Logging for 'dplyr' and 'tidyr' Functions

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

tidylog

The goal of tidylog is to provide feedback about dplyr and tidyr operations. It provides simple wrapper functions for almost all dplyr and tidyr functions, such as filter, mutate, select, full_join, and group_by.

Example

Load tidylog after dplyr and/or tidyr:

library("dplyr")
library("tidyr")
library("tidylog", warn.conflicts = FALSE)

Tidylog will give you feedback, for instance when filtering a data frame or adding a new variable:

filtered <- filter(mtcars, cyl == 4)
mutated <- mutate(mtcars, new_var = wt ** 2)

Tidylog reports detailed information for joins:

joined <- left_join(nycflights13::flights, nycflights13::weather,
    by = c("year", "month", "day", "origin", "hour", "time_hour"))

In this case, we see that 1,556 rows from the flights dataset do not have weather information.

Tidylog can be especially helpful in longer pipes:

summary <- mtcars %>%
    select(mpg, cyl, hp, am) %>%
    filter(mpg > 15) %>%
    mutate(mpg_round = round(mpg)) %>%
    group_by(cyl, mpg_round, am) %>%
    tally() %>%
    filter(n >= 1)

Here, it might have been accidental that the last filter command had no effect.

Installation

Download from CRAN:

install.packages("tidylog")

Or install the development version:

devtools::install_github("elbersb/tidylog")

Benchmarks

Tidylog will add a small overhead to each function call. This can be relevant for very large datasets and especially for joins. If you want to switch off tidylog for a single long-running command, simply prefix dplyr:: or tidyr::, such as in dplyr::left_join. See this vignette for more information.

More examples

filter, distinct, drop_na

a <- filter(mtcars, mpg > 20)
b <- filter(mtcars, mpg > 100)
c <- filter(mtcars, mpg > 0)
d <- filter_at(mtcars, vars(starts_with("d")), any_vars((. %% 2) == 0))
e <- distinct(mtcars)
f <- distinct_at(mtcars, vars(vs:carb))
g <- top_n(mtcars, 2, am)
i <- sample_frac(mtcars, 0.5)

j <- drop_na(airquality)
k <- drop_na(airquality, Ozone)

mutate, transmute, replace_na, fill

a <- mutate(mtcars, new_var = 1)
b <- mutate(mtcars, new_var = runif(n()))
c <- mutate(mtcars, new_var = NA)
d <- mutate_at(mtcars, vars(mpg, gear, drat), round)
e <- mutate(mtcars, am_factor = as.factor(am))
f <- mutate(mtcars, am = as.ordered(am))
g <- mutate(mtcars, am = ifelse(am == 1, NA, am))
h <- mutate(mtcars, am = recode(am, `0` = "zero", `1` = NA_character_))

i <- transmute(mtcars, mpg = mpg * 2, gear = gear + 1, new_var = vs + am)

j <- replace_na(airquality, list(Solar.R = 1))
k <- fill(airquality, Ozone)

joins

For joins, tidylog provides more detailed information. For any join, tidylog will show the number of rows that are only present in x (the first dataframe), only present in y (the second dataframe), and rows that have been matched. Numbers in parentheses indicate that these rows are not included in the result. Tidylog will also indicate whether any rows were duplicated (which is often unintentional):

x <- tibble(a = 1:2)
y <- tibble(a = c(1, 1, 2), b = 1:3) # 1 is duplicated
j <- left_join(x, y, by = "a")

More examples:

a <- left_join(band_members, band_instruments, by = "name")
b <- full_join(band_members, band_instruments, by = "name")
c <- anti_join(band_members, band_instruments, by = "name")

Because tidylog needs to perform two additional joins behind the scenes to report this information, the overhead will be larger than for the other tidylog functions (especially with large datasets).

select, relocate, rename

a <- select(mtcars, mpg, wt)
b <- select_if(mtcars, is.character)
c <- relocate(mtcars, hp)
d <- select(mtcars, a = wt, b = mpg)

e <- rename(mtcars, miles_per_gallon = mpg)
f <- rename_with(mtcars, toupper)

summarize

a <- mtcars %>%
    group_by(cyl, carb) %>%
    summarize(total_weight = sum(wt))

b <- iris %>%
    group_by(Species) %>%
    summarize_all(list(min, max))

tally, count, add_tally, add_count

a <- mtcars %>% group_by(gear, carb) %>% tally
b <- mtcars %>% group_by(gear, carb) %>% add_tally()

c <- mtcars %>% count(gear, carb)
d <- mtcars %>% add_count(gear, carb, name = "count")

pivot_longer, pivot_wider

longer <- mtcars %>%
    mutate(id = 1:n()) %>%
    pivot_longer(-id, names_to = "var", values_to = "value")
wider <- longer %>%
    pivot_wider(names_from = var, values_from = value)

Tidylog also supports gather and spread.

Turning logging off, registering additional loggers

To turn off the output for just a particular function call, you can simply call the dplyr and tidyr functions directly, e.g. dplyr::filter or tidyr::drop_na.

To turn off the output more permanently, set the global option tidylog.display to an empty list:

options("tidylog.display" = list())  # turn off
a <- filter(mtcars, mpg > 20)

options("tidylog.display" = NULL)    # turn on
a <- filter(mtcars, mpg > 20)

This option can also be used to register additional loggers. The option tidylog.display expects a list of functions. By default (when tidylog.display is set to NULL), tidylog will use the message function to display the output, but if you prefer a more colorful output, simply overwrite the option:

library("crayon")  # for terminal colors
crayon <- function(x) cat(red$bold(x), sep = "\n")
options("tidylog.display" = list(crayon))
a <- filter(mtcars, mpg > 20)

To print the output both to the screen and to a file, you could use:

log_to_file <- function(text) cat(text, file = "log.txt", sep = "\n", append = TRUE)
options("tidylog.display" = list(message, log_to_file))
a <- filter(mtcars, mpg > 20)

Namespace conflicts

Tidylog redefines several of the functions exported by dplyr and tidyr, so it should be loaded last, otherwise there will be no output. A more explicit way to resolve namespace conflicts is to use the conflicted package:

library("dplyr")
library("tidyr")
library("tidylog")
library("conflicted")
for (f in getNamespaceExports("tidylog")) {
    conflicted::conflict_prefer(f, "tidylog", quiet = TRUE)
}

elbersb/tidylog documentation built on Oct. 14, 2023, 12:54 p.m.