I have this type of data:

df <- structure(list(Utterance = c("(5.127)", ">like I don't understand< sorry like how old's your mom¿", 
                                   "(0.855)", "eh six:ty:::-one=", "(0.101)", "(0.487)", "[((v: gasps)) she said] ~no you're [not?]~", 
                                   "[((v: gasps)) she said] ~no you're [not?]~", "~<[NO YOU'RE] NOT (.) you can't go !in!>~", 
                                   "(0.260)", "show her [your boobs] next time"), 
                     Q = c(NA, "q_wh", "", "", NA, NA, "q_really", "", "", NA, NA), 
                     Sequ = c(NA, 1L, 1L, 1L, NA, NA, 0L, 0L, 0L, NA, NA)), class = "data.frame", row.names = c(NA, -11L))

I would like to extract/filter

• those rows where Sequ is not NA and
• the immediately preceding row (where Sequ is NA)

My attempt so far is to define a function that gets the indices of the relevant rows:

QA_sequ <- function(value) {
  inds <- which(!is.na(value) & lag(is.na(value)))  
  sort(unique(c(inds-1, inds)))
}

and then to slice out the rows via the indices:

library(dplyr)
df %>% 
  slice(QA_sequ(Sequ))
                                                 Utterance        Q Sequ
1                                                  (5.127)     <NA>   NA
2 >like I don't understand< sorry like how old's your mom¿     q_wh    1
3                                                  (0.487)     <NA>   NA
4               [((v: gasps)) she said] ~no you're [not?]~ q_really    0

However, only the immediately preceding row and first Sequ row are filtered. The result I want to obtain is this:

                                                  Utterance        Q Sequ
1                                                   (5.127)     <NA>   NA
2  >like I don't understand< sorry like how old's your mom¿     q_wh    1
3                                                   (0.855)             1
4                                         eh six:ty:::-one=             1
5                                                   (0.487)     <NA>   NA
6                [((v: gasps)) she said] ~no you're [not?]~ q_really    0
7                [((v: gasps)) she said] ~no you're [not?]~             0
8                 ~<[NO YOU'RE] NOT (.) you can't go !in!>~             0

EDIT:

The solution I've come up with feels cumbersome:

QA_sequ <- function(value) {
  inds <- which(!is.na(value) & lag(is.na(value)))  
  sort(unique(c(inds-1)))    # extract only preceding row!
}

library(dplyr)
df %>% 
  mutate(id = row_number()) %>%
  slice(QA_sequ(Sequ)) %>%
  bind_rows(., df %>% mutate(id = row_number()) %>% filter(!is.na(Sequ))) %>%
  arrange(id)

CodePudding user response：

How about this?

df %>%
  filter(!is.na(Sequ) | lead(!is.na(Sequ), default=FALSE))
#                                                  Utterance        Q Sequ
# 1                                                  (5.127)     <NA>   NA
# 2 >like I don't understand< sorry like how old's your mom¿     q_wh    1
# 3                                                  (0.855)             1
# 4                                        eh six:ty:::-one=             1
# 5                                                  (0.487)     <NA>   NA
# 6               [((v: gasps)) she said] ~no you're [not?]~ q_really    0
# 7               [((v: gasps)) she said] ~no you're [not?]~             0
# 8                ~<[NO YOU'RE] NOT (.) you can't go !in!>~             0

The logic filters (extracts) both of:

• all non-NA values
• any NA value where the next value is not NA

CodePudding user response：

Just add an additional OR to collect the rows where sequ is not NA but which don't have a corresponding lagged non-NA...

QA_sequ <- function(value) {
  inds <- which((!is.na(value) & lag(is.na(value))) | !is.na(value))  
  sort(unique(c(inds-1, inds)))
}

df %>%  slice(QA_sequ(Sequ))
                                                 Utterance        Q Sequ
1                                                  (5.127)     <NA>   NA
2 >like I don't understand< sorry like how old's your mom¿     q_wh    1
3                                                  (0.855)             1
4                                        eh six:ty:::-one=             1
5                                                  (0.487)     <NA>   NA
6               [((v: gasps)) she said] ~no you're [not?]~ q_really    0
7               [((v: gasps)) she said] ~no you're [not?]~             0
8                ~<[NO YOU'RE] NOT (.) you can't go !in!>~             0

CodePudding user response：

This is using base R. Just as you were doing take the indices of rows then the preceding.

x<-which(!is.na((df$Sequ))) 
x1 <- x-1
x<- unique(c(x,x1))
x<- x[order(x)]    


df[x,]

You can pass the same vector into slice to df %>% slice(x)