Introduction to R for Data Science :: Session 5 [Data Structuring: Strings in R]
4 likes14,391 views
This document provides an introduction and overview of working with strings in R. It covers the base string functions as well as functions from the stringr and stringi packages. Topics covered include concatenation, splitting, subsetting, transforming, searching, sorting, and encodings when working with strings in R. The document is presented as notes from a session on structuring data with strings in R as part of an introduction to R for data science course.
![Intro to R for Data Science
Session 5: Structuring Data: Strings in R
# Introduction to R for Data Science
# SESSION 5 :: 26 May, 2016
stringA <- "The quick brown fox jumps overthe lazy dog";
splitA <- strsplit(stringA," ") # is.list(splitA) == T
splitA <- unlist(strsplit(stringA," "))
# "The quick brown" from "The quick brown fox jumps overthe lazy dog"
splitA <- paste(unlist(strsplit(stringA," "))[1:3],collapse=" ")
# or
splitA <- paste(strsplit(stringA," ")[[1]][1:3],collapse=" ")
# advice:use
splitA <- strsplit(stringA," ",fixed=T) # fixed=T says:match the split argumentexactly,
# otherwise,split is an regularexpression;defaultis: fixed = FALSE
# string split w. {stringr}
is.list(str_split(stringA," "))
# this is interesting:
str_split(stringA," ", n=3)
# "The quick brown" from "The quick brown fox jumps overthe lazy dog"
paste0(str_split(stringA," ", n=4)[[1]][1:3],collapse=" ")
Strings in R
• Splitting](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-4-320.jpg)
![Intro to R for Data Science
Session 5: Structuring Data: Strings in R
# Introduction to R for Data Science
# SESSION 5 :: 26 May, 2016
# default: str_split(string,pattern,n = Inf), where pattern is regex
str_split(stringA,boundary("word"))
# very useful:
stringA1 <- "The quick brown fox jumps overthe lazy dog"
str_split(stringA1,boundary("word"))
stringA1 <- "Aboveall, don'tlie to yourself.
The man who lies to himselfand listens to his own lie comes to a pointthat he cannotdistinguish the
truth within him, or around him,and so loses all respectfor himselfand for others.
And having no respecthe ceasesto love."
str_split(stringA1,boundary("word"))
str_split(stringA1,boundary("word",skip_word_none= F)) # includingpunctuation and special
str_split(stringA1,boundary("line_break"))
writeLines(str_split(stringA1,boundary("line_break"))[[1]])
Strings in R
• Splitting](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-5-320.jpg)
![Intro to R for Data Science
Session 5: Structuring Data: Strings in R
# Introduction to R for Data Science
# SESSION 5 :: 26 May, 2016
# Searchingfor somethingin a string
# Does a string encompass a substring?
grepl("Thequick brown",stringA,fixed = T)
grepl("Thefastred", stringA, fixed = T)
stringB <- "Uraaaaaaaa"
grep("Thequick brown",c(stringA,stringB),fixed = T)
# where?
stringA <- "The quick brown fox jumps overthe lazy dog The quick brown"
w <- gregexpr("Thequick brown",stringA)
str(w)
b1 <- w[[1]][1] # first match starts at
b2 <- w[[1]][2] # second match starts at
# now, match.length is an attribute of w[[1]], not w itself:
e1 <- attr(w[[1]],"match.length",exact= T)[1]
e2 <- attr(w[[1]],"match.length",exact= T)[2]
Strings in R
• Search](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-9-320.jpg)
![Intro to R for Data Science
Session 5: Structuring Data: Strings in R
# Introduction to R for Data Science
# SESSION 5 :: 26 May, 2016
# first match extraction:
str_sub(stringA,b1,b1+e1-1)
# second matchextraction:
str_sub(stringA,b2,b2+e2-1)
# Ok, but easierand more convenientwith {stringr}
str_detect(stringA,"The quickbrown") # T or F
str_locate(stringA,"The quickbrown") # first match
str_locate_all(stringA,"The quickbrown") # all matches
# term frequency,as we know,is very importantin text-mining:
term1 <- str_locate_all(stringA,"The quickbrown")[[1]]# all matches for term1 ie. "The quick
brown"
dim(term1)[1] # how many matches = how many rows in the str_locate_alloutputmatrix...
Strings in R
• Search](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-10-320.jpg)
![Intro to R for Data Science
Session 5: Structuring Data: Strings in R
# Introduction to R for Data Science
# SESSION 5 :: 26 May, 2016
# Sorting strings in R
letters
str_sort(letters,locale="en")# locale = en
str_sort(letters,locale="haw")# locale = Hawaiian
# backwards
str_sort(letters,decreasing= T)
# handy:
stringA <- c("New York","Paris",NA, "Moscow","Tokyo")
str_sort(stringA,na_last=T)
# [1] "Moscow" "New York" "Paris" "Tokyo" NA
str_sort(stringA,na_last=F)
# [1] NA "Moscow" "New York" "Paris" "Tokyo"
# {base}
sort(stringA)
sort(stringA,decreasing=T)
Strings in R
• Sorting strings](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-11-320.jpg)
![Introduction to R for Data Science :: Session 5 [Data Structuring: Strings in R]](https://image.slidesharecdn.com/intrordatasciencesession5eng-160529191417/85/Introduction-to-R-for-Data-Science-Session-5-Data-Structuring-Strings-in-R-13-320.jpg)
Introduction to R for Data Science :: Session 5 [Data Structuring: Strings in R]
- 1. Introduction to R for Data Science Lecturers dipl. ing Branko Kovač Data Analyst at CUBE/Data Science Mentor at Springboard Data Science zajednica Srbije [email protected] dr Goran S. Milovanović Data Scientist at DiploFoundation Data Science zajednica Srbije [email protected] [email protected]
- 2. Strings in R • {base} for strings • {stringr} for strings • {stringi} for strings Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Processing strings in R library(stringr) # strings in R are charactervectors stringA <- "Hello world" stringB <- "Sun shines!" stringA stringB is.character(stringA) # TRUE as.character(200*5) as.numeric("1000") as.double("3.14") # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Using " and ' # either: stringA <- "Hello 'World'" stringA # or stringA <- 'Hello "World"' stringA # prints:"Hello "World"" - what is this: ? print(stringA) # try: writeLines(stringA) print(stringA) # Escapingin R: use , the R escape character stringA <- 'Hello "World"' stringA print(stringA) writeLines(stringA) # Escapingescaping writeLines("")# nice
- 3. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # String Concatenationin R stringC <- c(stringA,stringB) # a character vectorof length == 2 length(stringC) stringC <- paste(stringA,stringB, sep=",") # length == 1, base function writeLines(stringC) # sep w. collapse (paste args) stringC <- c(stringA,stringB) stringC <- paste(stringC,collapse="__") writeLines(stringC) # paste0 is paste w. sep="",fasterthan paste(),base function # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 strA <- "One" strB <- "Two" strC <- "Three" paste0(strA,strB, strC) # the collapse argumentis used in paste0 as well strD <- c(strA,strB,strC) paste0(strD,collapse="-") # stringr concatenation,also has sep and collapse as args str_c(strA,strB,strC) str_c(strA,strB,strC,sep="...") str_c(strD,collapse="...") # both paste {base}and str_c {stringr} are vectorized paste("Prefix-",strD, sep="-") str_c("Prefix-",strD,sep="-") Strings in R • Concatenation
- 4. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 stringA <- "The quick brown fox jumps overthe lazy dog"; splitA <- strsplit(stringA," ") # is.list(splitA) == T splitA <- unlist(strsplit(stringA," ")) # "The quick brown" from "The quick brown fox jumps overthe lazy dog" splitA <- paste(unlist(strsplit(stringA," "))[1:3],collapse=" ") # or splitA <- paste(strsplit(stringA," ")[[1]][1:3],collapse=" ") # advice:use splitA <- strsplit(stringA," ",fixed=T) # fixed=T says:match the split argumentexactly, # otherwise,split is an regularexpression;defaultis: fixed = FALSE # string split w. {stringr} is.list(str_split(stringA," ")) # this is interesting: str_split(stringA," ", n=3) # "The quick brown" from "The quick brown fox jumps overthe lazy dog" paste0(str_split(stringA," ", n=4)[[1]][1:3],collapse=" ") Strings in R • Splitting
- 5. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # default: str_split(string,pattern,n = Inf), where pattern is regex str_split(stringA,boundary("word")) # very useful: stringA1 <- "The quick brown fox jumps overthe lazy dog" str_split(stringA1,boundary("word")) stringA1 <- "Aboveall, don'tlie to yourself. The man who lies to himselfand listens to his own lie comes to a pointthat he cannotdistinguish the truth within him, or around him,and so loses all respectfor himselfand for others. And having no respecthe ceasesto love." str_split(stringA1,boundary("word")) str_split(stringA1,boundary("word",skip_word_none= F)) # includingpunctuation and special str_split(stringA1,boundary("line_break")) writeLines(str_split(stringA1,boundary("line_break"))[[1]]) Strings in R • Splitting
- 6. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 stringA <- c("Belgrade","Zagreb","Ljubljana")# {stringr} str_sub(stringA,1, 2) # counting backwards str_sub(stringA, -3, -1) # {base} substr(stringA,1, 3) # play: substr(stringA,c(1,2,3),c(2,3,4)) # nope: substr(stringA, -2, -1) # {base} Strings in R • Subsetting strings
- 7. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Replacingcharactersin strings stringB <- stringA # just a copy of stringA str_sub(stringB,1,2)<- "00" stringB # {base} stringB <- stringA # just a copy of stringA substr(stringB,1,3)<- "WowWow" # check the result! stringB substr(stringB,1,4)<- "WoWWow" # check the result! stringB substr(stringB,1,6)<- "WowWow" # check the result! stringB Strings in R • Subsetting strings # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # UPPER CASE to lower case and vice versa in R stringA <- "ABRACADABRA" # {base} tolower(stringA) stringA <- tolower(stringA) toupper(stringA) stringA <- toupper(stringA) # {stringr} str_to_lower(stringA) stringB <- str_to_lower(stringA) str_to_upper(stringA) # capitalize first letter str_to_title(stringB) • Transforming strings
- 8. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Remove whitespace stringA <- c(" Removewhitespace "); str_trim(stringA) # remove leading whitespace str_trim(stringA,side="left") # remove trailing whitespace str_trim(stringA,side="right") # remove all whitespace? stringA <- c(" Remove whitespace ") # how aboutthis one? # there are differentways to do it. Try: gsub(" ", "", stringA,fixed=T) # (!(fixed==T)),the first (pattern) argumentis regex # in general: stringA <- "The quick brown fox jumps overthe lazy dog The quick brown" gsub("Thequick brown","The slow red", stringA,fixed=T) Strings in R • More transforming
- 9. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Searchingfor somethingin a string # Does a string encompass a substring? grepl("Thequick brown",stringA,fixed = T) grepl("Thefastred", stringA, fixed = T) stringB <- "Uraaaaaaaa" grep("Thequick brown",c(stringA,stringB),fixed = T) # where? stringA <- "The quick brown fox jumps overthe lazy dog The quick brown" w <- gregexpr("Thequick brown",stringA) str(w) b1 <- w[[1]][1] # first match starts at b2 <- w[[1]][2] # second match starts at # now, match.length is an attribute of w[[1]], not w itself: e1 <- attr(w[[1]],"match.length",exact= T)[1] e2 <- attr(w[[1]],"match.length",exact= T)[2] Strings in R • Search
- 10. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # first match extraction: str_sub(stringA,b1,b1+e1-1) # second matchextraction: str_sub(stringA,b2,b2+e2-1) # Ok, but easierand more convenientwith {stringr} str_detect(stringA,"The quickbrown") # T or F str_locate(stringA,"The quickbrown") # first match str_locate_all(stringA,"The quickbrown") # all matches # term frequency,as we know,is very importantin text-mining: term1 <- str_locate_all(stringA,"The quickbrown")[[1]]# all matches for term1 ie. "The quick brown" dim(term1)[1] # how many matches = how many rows in the str_locate_alloutputmatrix... Strings in R • Search
- 11. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Sorting strings in R letters str_sort(letters,locale="en")# locale = en str_sort(letters,locale="haw")# locale = Hawaiian # backwards str_sort(letters,decreasing= T) # handy: stringA <- c("New York","Paris",NA, "Moscow","Tokyo") str_sort(stringA,na_last=T) # [1] "Moscow" "New York" "Paris" "Tokyo" NA str_sort(stringA,na_last=F) # [1] NA "Moscow" "New York" "Paris" "Tokyo" # {base} sort(stringA) sort(stringA,decreasing=T) Strings in R • Sorting strings
- 12. Intro to R for Data Science Session 5: Structuring Data: Strings in R # Introduction to R for Data Science # SESSION 5 :: 26 May, 2016 # Take home messageon encodings # 1. Most of the time, you simply need to know the source encoding # 2. All of the time *** converteverythingto UTF-8*** - as soon as possible # 3. Most {base},and all {stringr} and {stringi} functions thatprocessstrings in R # will converttheir outputto UTF-8 automatically # Working inside R only, running an English locale,will nevercause you any trouble # However,in Data Science you will probably needto do a lot of web-scraping fora living # - and that's where the fan starts. # God bless iconv()- but don'tget to excited,it does not avoid all problems # Next session:Thurday,June2, 2016 :: LinearRegressionw. R Strings in R • Encodings…