CODEWITHSUNDEEP
rdata.table
user3122260
user3122260

Reputation: 53

selecting consecutive answers in R

I have data set as follows (it is just a sample below):

dataframe<-data.frame("id" = c(1,2,5,7,9,21,22,23),"questionfk"=c(145,51,51,145,145,51,145,51))

In this data id represents the order of the questions. Questionfk, is the question id.

I would like to filter this data on questionfk 145 and 51, where 145 is asked right before 51 was the second question after. So what I want in the end seems like below:

dataframefiltered<-data.frame("id" = c(1,2,22,23),"questionfk"=c(145,51,145,51))

I did this with lots of if's and for's is it possible to do this with data.table? and How?

Thank you!

Upvotes: 0

Views: 56

Answers (2)

akrun
akrun

Reputation: 887078

May be this helps

library(data.table)
setDT(dataframe)[dataframe[, {indx=which(c(TRUE, questionfk[-1]==145 &
         questionfk[-.N]==51) & c(TRUE, diff(id)==1))
          sort(c(indx, indx+1))}]]
#    id questionfk
#1:  1        145
#2:  2         51
#3: 22        145
#4: 23         51

Upvotes: 1

eamcvey
eamcvey

Reputation: 693

I'm not sure I understand the exact conditions you're looking for, but I'm basing this on wanting to select questions 145 and 51, but only when then come consecutively in that order. I realize that this does not give the same result as you show, but presumably you can modify this to match the right conditions.

Rather than data.table, here's a way to do it with dplyr (which is also fast with big datasets, and very elegant):

dataframe %>% 
  mutate(last_question = lag(questionfk),
         next_question = lead(questionfk),
         after_145 = last_question==145,
         before_51 = next_question==51) %>%
  filter(after_145 | before_51) %>%
  select(id, questionfk)

Upvotes: 0

Related Questions