12

I have a data frame in the form of columns - input

Id  Comment
xc545   Ronald is a great person 
g6548   Hero worship is bad

I need the output in the form of Result

Id  Words 
xc545   Ronald
xc545   is
xc545   a
xc545   great
xc545   person
g6548   Hero
g6548   worship
g6548   is
g6548   bad

Need a R statement to execute this.

Following is what I tried - 

result<-lapply(input,function(x)strsplit(x[2]," "))

However this returns only one record.

agstudy
  • 119,832
  • 17
  • 199
  • 261
user1882633
  • 153
  • 1
  • 2
  • 9
  • 1
    May I ask why you want this? That may determine if a dataframe is the best structure for your data. – Tyler Rinker Jun 04 '13 at 15:30
  • I wanted to search across the sentence for productnames and tag them. I have the productnames(200,000) in another table column. SQL comparison is taking very long. – user1882633 Jun 04 '13 at 15:53

3 Answers3

9

Suppose DF is your data.frame, a possibility could be:

> List <- strsplit(DF$Comment, " ")
> data.frame(Id=rep(DF$Id, sapply(List, length)), Words=unlist(List))
     Id   Words
1 xc545  Ronald
2 xc545      is
3 xc545       a
4 xc545   great
5 xc545  person
6 g6548    Hero
7 g6548 worship
8 g6548      is
9 g6548     bad

Note that my answer is only valid when there is a simple space between each pair of words.

Jilber Urbina
  • 58,147
  • 10
  • 114
  • 138
4

A data.table solution inspired from this one:

library(data.table)
dt = data.table(df)
dt[,c(Words=strsplit(Comment, " ", fixed = TRUE)), by = Id]
Id      V1
1: xc545  Ronald
2: xc545      is
3: xc545       a
4: xc545   great
5: xc545  person
6: g6548    Hero
7: g6548 worship
8: g6548      is
9: g6548     bad
Community
  • 1
  • 1
agstudy
  • 119,832
  • 17
  • 199
  • 261
3

Using scan, tapply and stack:

d <- read.table(text='Id  Comment
xc545   "Ronald is a great person"
g6548   "Hero worship is bad"', header=TRUE, as.is=TRUE)

stack(tapply(d$Comment, d$Id, function(x) scan(text=x, what='')))
#    values   ind
# 1    Hero g6548
# 2 worship g6548
# 3      is g6548
# 4     bad g6548
# 5  Ronald xc545
# 6      is xc545
# 7       a xc545
# 8   great xc545
# 9  person xc545
Matthew Plourde
  • 43,932
  • 7
  • 96
  • 113