The ultimate R data.table cheat sheet

How do I do that in data.table? Find code for dozens of data tasks in this searchable cheat sheet of R data.table and Tidyverse code.

Executive Editor, Data & Analytics, InfoWorld |

The ultimate R data.table cheat sheet — Thinkstock

Speed. Concise syntax. Backwards compatibility.

But especially speed.

Those are some of the features that make R’s data.table compelling to its fans.

And even if you’re a confirmed tidyverse user (as I am), data.table can be a useful addition to your R toolkit — especially when working with large data sets, in a Shiny app, or in a package function.

This R data.table ultimate cheat sheet is different from many others because it’s interactive. You can search for a specific phrase like add column or by a type of task group such as Subset or Reshape. In addition, because this cheat sheet includes tidyverse “translation” code, you can also search for a favorite dplyr verb such as mutate or rowwise.

Registered InfoWorld users can download an expanded version of this interactive table for use on their own systems! Registration is free.

Search data.table and tidyverse tasks and code

Task	Type	data.table Code	Tidyverse Code
Read in CSV file	Import	mydt <- fread("myfile.csv")	myt <- read_csv("myfile.csv") #OR myt <- vroom::vroom("myfile.csv")
Import the first x number of rows of a CSV file	Import	mydt_x <- fread("myfile.csv", nrows = x)	myt_x <- read_csv("myfile.csv", n_max = x)
Import only those rows from a CSV file that match a certain pattern	Import	mydt_pattern <- fread("grep 'mypattern' myfile.csv")	myt_pattern <- vroom::vroom(pipe("grep 'mypattern' myfile.csv"))
Import a .gz compressed file	Import	mydt <- fread("myfile.gz")	myt <- vroom::vroom("myfile.gz")
Import a.zip compressed file	import	mydt <- fread(cmd = 'unzip -cq myfile.zip')	myt <- read_csv("myfile.zip")
Create data table from existing data frame (tibble for tidyverse)	Import	mydt <- as.data.table(mydf) #OR setDT(mydf)	myt <- as_tibble(mydf)
Alter data.table in place without making a copy	Wrangle	any function that starts with set such as setkey(mydt, mycol) or using the := operator within brackets	not applicable
Order rows based on multiple column values	Wrangle	mydt2 <- mydt[order(colA, -colB)] #OR setorder(mydt, colA, -colB)	myt <- arrange(myt, colA, -colB)
Rename columns	Wrangle	setnames(mydt, old = c('colA','colB'), new = c('NewColA', 'NewColB'))	myt <- rename(myt, NewColA = colA, NewColB = colB)
Reordering columns: Move some columns to the front (left-most) position	Wrangle	setcolorder(mydt, c("colB", "colC")) # colB now in position 1 and colC in position 2	myt <- relocate(myt, colB, colC)
Filter rows for row number n	Subset	mydt2 <- mydt[n]	myt2 <- slice(myt, n)
Filter for the last row	Subset	mydt2 <- mydt[.N]	myt2 <- slice(myt, n())
Filter rows by condition	Subset	# In some cases setkey(mydt, colA, colB) will speed performance # for logical tests on colA and colB; same with other columns mydt2 <- mydt[logical expression]	myt2 <- filter(myt, logical expression)
Filter rows where colA equals string1 or string2	Subset	mydt2 <- mydt[colA %chin% c("string1", "string2")]	myt2 <- filter(myt, colA %in% c("string1", "string2"))
Filter rows where colA matches a regular expression	Subset	mydt2 <- mydt[colA %like% "mypattern"]	myt2 <- filter(myt, stringr::str_detect(colA, "mypattern"))
Filter rows where colA values are between 2 numbers	Subset	mydt2 <- mydt[colA %between% c(n1, n2)]	myt2 <- filter(myt, between(colA, n1, n2))
Filter for first n rows by group	Subset	mydt2 <- mydt[, .SD[1:n], by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% slice(1:n)
Filter rows for maximum value by group	Subset	mydt2 <- mydt[, .SD[which.max(valcol)], by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% filter(valcol == max(valcol))
Select column and return results as a vector	Subset	myvec <- mydt[, colname]	myvec <- pull(myt, colname)
Select multiple columns to create a new data.table (data frame or tibble for tidyverse)	Subset	mydt2 <- mydt[, list(colA, colB)] #OR mydt2 <- mydt[, .(colA, colB)] #OR mydt2 <- mydt[, c("colA", "colB")]	myt2 <- select(myt, colA, colB)
Select multiple columns using a variable containing the column names	Subset	my_col_names <- c("colA", "colB") mydt2 <- mydt[, ..my_col_names] #OR mydt2 <- mydt[, my_col_names, with = FALSE]	my_col_names <- c("colA", "colB") myt2 <- select(myt, all_of(my_col_names))
Select multiple columns and rename some	Subset	mydt2 <- mydt[, .(newname1 = col1, newname2 = col2, col3)]	myt2 <- select(myt, newname1 = col1, newname2 = col2, col3)
Exclude multiple columns	Subset	mydt2 <- mydt[, -c("colA", "colB")] #OR mydt2 <- mydt[, !c("colA", "colB")] #OR my_col_names <- c("colA", "colB") mydt2 <- mydt[, !..my_col_names]	myt2 <- select(myt, -c(colA, colB)) #OR my_col_names <- c("colA", "colB") myt2 <- select(myt, -{{my_col_names}})
Remove duplicate rows based on values in multiple columns	Subset	mydt2 <- unique(mydt, by = c("colA", "colB"))	myt2 <- distinct(myt, colA, colB, .keep_all = TRUE)
Count unique rows based on multiple columns	Summarize	uniqueN(mydt, by = c("colA", "colB"))	nrow(distinct(myt, colA, colB))
Run summary calculations on data	Summarize	mydt2 <- mydt[, myfun(colA ...)]	myt2 <- myt %>% summarise( ColName = myfun(colA ...) )
Run summary calculations on data by one group	Summarize	mydt2 <- mydt[, myfun(colA ...), by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% summarise( NewCol = myfun(colA...) )
Run summary calculations on data by one group and name new column	Summarize	mydt2 <- mydt[, .(MyNewCol = myfun(colA...)), by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% summarise( NewCol = myfun(colA...) )
Run summary calculations on data by multiple groups	Summarize	mydt2 <- mydt[, myfun(colA ...), by = .(groupcol1, groupcol2)]	myt2 <- myt %>% group_by(groupcol1, groupcol2) %>% summarise( NewCol = myfun(colA...) )
Run summary calculation on filtered data by multiple groups	Summarize	mydt2 <- mydt[filter expression, myfun(colA), by = .(groupcol1, groupcol2)]	myt2 <- myt %>% filter(filter expression) %>% group_by(groupcol1, groupcol2) %>% summarise( NewCol = myfun(colA), .groups = "keep" )
Count number of rows by groups	Summarize	mydt2 <- mydt[,.N, by = groupcol] #for one group #OR mydt2 <- mydt[, .N, by = .(groupcol1, groupcol2)]	myt2 <- count(myt, groupcol) #for one group #OR myt2 <- count(myt, groupcol1, groupcol2)
Summarize multiple columns and return results in multiple columns	Summarize	mydt2 <- mydt[, lapply(.SD, myfun), .SDcols = c("colA", "colB")]	myt2 <- myt %>% summarise( across(c(colA, colB), myfun) )
Summarize multiple columns by group and return results in multiple columns	Summarize	mydt2 <- mydt[, lapply(.SD, myfun), .SDcols = c("colA", "colB"), by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% summarise( across(c(colA, colB), myfun) )
Add a column	Calculate	mydt[, MyNewCol := myfun(colA)]	myt <- myt %>% mutate( MyNewCol = myfun(colA) )
Add multiple columns at once	Calculate	# use any function or expression mydt[, `:=`(NewCol1 = myfun(colA), NewCol2 = colB + colC )] #OR mydt[, c("NewCol1", "newCol2") := list(myfun(colA), colB + colC)]	myt <- myt %>% mutate( MyNewCol1 = myfun(colA), MyNewCol2 = colB + colC )
Add column using current and previous values from another column, such as finding the difference between value on a date vs. the prior date	Calculate	mydt[, Diff := colA - shift(colA)]	myt <- mutate(myt, Diff = colA - lag(colA))
Add column referencing previous value of a column by a group	Calculate	mydt2 <- mydt[, Diff := colA - shift(colA), by = groupcol]	myt2 <- myt %>% group_by(groupcol) %>% mutate( Diff = colA - lag(colA) )
Add column with row ID numbers by group	Calculate	mydt[, myid := 1:.N, by = groupcol]	myt <- myt %>% group_by(groupcol) %>% mutate( myid = row_number() )
Add column based on several conditions without using multiple if else statements (like SQL's CASE)	Calculate	# Needs data.table version 1.13 or later # I like each condition on a new line but that's not required mydt2 <- mydt[, NewCol := fcase( condition1, "Value1", condition2, "Value2", condition3, "Value3", default = "Other" # value for all else )]	myt2 <- myt %>% mutate( NewCol = case_when( condition1 ~ "Value1", condition2 ~ "Value2", condition3 ~ "Value3", TRUE ~ "Other" ) )
Add column via operating by row	Calculate	mydt[, newcol := myfun(colB, colC, colD), by = 1:nrow(mydt)] # or if colA has all unique values mydt[, newcol := myfun(colB, colC, colD), by = colA]	myt <- myt %>% rowwise() %>% mutate( newcol = myfun(colB, colC, colD) ) # or myt <- myt %>% rowwise() %>% mutate( #use dplyr select syntax: newcol = myfun(c_across(colB:colD)) )
Join two data sets by more than one column; keep all in set1 but only matches in set2	Join	mydt <- dt2[dt1, on = c("dt2col" = "dt1col")] #OR mydt <- merge(dt1, dt2, by.x = "dt1col", by.y = "dt2col", all.x = TRUE) #OR setkey(dt1, "dt1col") setkey(dt2, "dt2col") mydt <- dt2[dt1]	myt <- left_join(df1, df2, by = c("df1col" = "df2col"))
Join 2 data sets by more than one column - keep all in set1 but only matches in set2	Join	mydt <- merge(dt1, dt2, by.x = c("dt1colA", "dt1colB"), by.y = c("dt2colA", "dt2colB"), all.x = TRUE, all.y = FALSE) #OR setkey(dt1, dt1colA, dt1colB) setkey(dt2, dt2colA, dt2colB) mydt <- dt2[dt1]	myt <- left_join(df1, df2, by = c("df1colA" = "df2colA", "df1colB" = "df2colB"))
Join two data sets by one common column; only keep matches	Join	mydt <- merge(dt1, dt2, by.x = "dtcol1", by.y = "dtcol2")	myt <- inner_join(df1, df2, by = c("df1col" = "df2col"))
Join two data sets by one common column and keep all data in both sets, whether or not there are matches	Join	mydt <- merge(dt1, dt2, by.x = "dtcol1", by.y = "dtcol2", all = TRUE)	myt <- full_join(df1, df2, by = c("df1col" = "df2col"))
Combine two data sets by adding rows from one to the bottom of another	Join	mydt_joined <- rbindlist(list(mydt, mydt2))	myt_joined <- bind_rows(myt, myt2)
Reshape data wide to long	Reshape	mydt_long <- melt(mydt, measure.vars = c("col1", "col2", "col3"), variable.name = "NewCategoryColName", value.name = "NewValueColName")	myt_long <- pivot_longer(myt, cols = starts_with("col"), names_to = "NewCategoryColName", values_to = "NewValueColName")
Reshape data long to wide	Reshape	mydt_wide <- dcast(mydt, id_col1 ~ col1 , value.var = "ValueColName")	myt_wide <- pivot_wider(myt, names_from = col1, values_from =ValueColName)
Chain multiple expressions	Wrangle	mydt[expr1][expr2]	myt <- myt %>% expr1 %>% expr2
Export data to a CSV file	Export	fwrite(mydt, "myfile.csv")	write_csv(myt, "myfile.csv")
Append rows to an existing CSV file	Export	fwrite(mydt2, "myfile.csv", append = TRUE)	vroom::vroom_write(myt2, "myfile.csv", delim = ",", append = TRUE)
Export data to a compressed CSV file	Export	fwrite(mydt, "myfile.csv.gz", compress = "gzip")	vroom::vroom_write(myt, "myfile2.csv.gz")

There is a lot more to learn about data.table! For some data.table basics, check out my five-minute introductory video:

Finally, the data.table website has lots more information and tips, such as using setkey() and other indexing tips.

Next read this:

Sharon Machlis is Director of Editorial Data & Analytics at Foundry, where she works on data analysis and in-house editor tools in addition to writing. Her book Practical R for Mass Communication and Journalism was published by CRC Press. She was named Digital Analytics Association's 2021 Top (Data) Practitioner, winner of the 2023 Jesse H. Neal journalism award for best instructional content, 2014 Azbee national gold award for investigative reporting, and 2017 Azbee gold for how-to article, among other awards. You can find her on Mastodon at masto.machlis.com/@smach.