Subset dataframe in R, dplyr filter row values of column A not NA in row of column B

Question

I have a dataset consisting of a time series study. Since some participants didn't show up for certain days, they have NA values for rest of the data frame, but certain study days were crucial, so I am trying to subset my data to participants not missing these crucial days. My dataset is actually very large but here's the general structure:

fakedat <- data.frame(ID = c("A", "A", "A", "A", "B", "B", "B", "B", "C", "C", "C", "C", 
                          "D", "D", "D", "D", "E", "E", "E", "E", "F", "F", "F", "F"),
                           StudyDay = c(1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4, 
                                        1, 2, 3, 4),
                           Ab = c(10, NA, 15, 10, 10, 20, 10, NA, 10, 10, NA, 30, NA, NA, 15, NA, 10, 20,
                                  10, 30, NA, 10, NA, 20))

Now let's say it was crucial they show up at day 2 and 4, I tried subsetting using dplyr filtering like this:

fakedat2 <- fakedat %>%
  dplyr::group_by(ID) %>%
  dplyr::filter(StudyDay %in% c(2, 4) & !is.na(Ab)) %>%
  dplyr:: ungroup()

EDIT: But the output of this dataset is only the list if IDs that have a 2 or 4 that's not an NA value. I need to find (in my real data) subjects who have NA Ab values at 4 specific Study Days. The answer I accepted below works but still curious about performing conditional filtering? Like in SAS you could code "IF Ab.=NA at (StudyDay=2 AND StudyDay=4) THEN ID....or something like that.

Answer 1

Maybe this will achieve your goal. If all participants have all StudyDay timepoints, and you just want to see if not missing in days 2 or 4, you can just check the Ab values at those time points in your filter . In this case, an ID will be omitted if is NA in both days 2 and 4 (in this example, "D").

Alternatively, if you want to require that both values are available for days 2 and 4, you can use & (AND) instead of | (OR).

library(dplyr)

fakedat %>%
  group_by(ID) %>%
  filter(!is.na(Ab[StudyDay == 2]) | !is.na(Ab[StudyDay == 4]))

If you have multiple days to check are not missing, you can use all and check values for NA where the StudyDay is %in% a vector of required days as follows:

required_vals <- c(2, 4)

fakedat %>%
  group_by(ID) %>%
  filter(all(!is.na(Ab[StudyDay %in% required_vals])))

Output

   ID    StudyDay    Ab
   <chr>    <dbl> <dbl>
 1 A            1    10
 2 A            2    NA
 3 A            3    15
 4 A            4    10
 5 B            1    10
 6 B            2    20
 7 B            3    10
 8 B            4    NA
 9 C            1    10
10 C            2    10
11 C            3    NA
12 C            4    30
13 E            1    10
14 E            2    20
15 E            3    10
16 E            4    30
17 F            1    NA
18 F            2    10
19 F            3    NA
20 F            4    20

Answer 2

In base R , we can do

subset(fakedat, ID %in% ID[StudyDay %in% c(2, 4) & !is.na(Ab)])

-output

#    ID StudyDay Ab
#1   A        1 10
#2   A        2 NA
#3   A        3 15
#4   A        4 10
#5   B        1 10
#6   B        2 20
#7   B        3 10
#8   B        4 NA
#9   C        1 10
#10  C        2 10
#11  C        3 NA
#12  C        4 30
#17  E        1 10
#18  E        2 20
#19  E        3 10
#20  E        4 30
#21  F        1 NA
#22  F        2 10
#23  F        3 NA
#24  F        4 20

---

Or a similar option in dplyr

library(dplyr)
fakedat %>%
     filter(ID %in% ID[StudyDay %in% c(2, 4) & !is.na(Ab)])