Counting number of rows between rows

Question

I have a data frame with four columns: user_id, event, and time

User_id a user_id, event is either \"A\" or \"B\", and time is time. I need to count the number of \"B\"

Answer 1

x <- read.table(header=TRUE, stringsAsFactors=FALSE, text='
user_id   event   date_time    desired_column
1         B       2018-01-01   NA
1         B       2018-01-02   NA
1         B       2018-01-03   NA
1         B       2018-01-04   NA
1         B       2018-01-05   NA
1         A       2018-01-06   5
1         B       2018-01-07   NA
1         A       2018-01-08   1
2         B       2018-01-05   NA
2         B       2018-01-06   NA
2         A       2018-01-07   2')

Perhaps a little clunky, but ...

(edit: specified dplyr::lag, since stats::lag doesn't do what we need.)

x$a <- NA
x$a[cumsum(rle(x$event)$lengths)] <- rle(x$event)$lengths
x$a <- dplyr::lag(x$a)
x$a[x$event == "B"] <- NA

x
#    user_id event  date_time desired_column  a
# 1        1     B 2018-01-01             NA NA
# 2        1     B 2018-01-02             NA NA
# 3        1     B 2018-01-03             NA NA
# 4        1     B 2018-01-04             NA NA
# 5        1     B 2018-01-05             NA NA
# 6        1     A 2018-01-06              5  5
# 7        1     B 2018-01-07             NA NA
# 8        1     A 2018-01-08              1  1
# 9        2     B 2018-01-05             NA NA
# 10       2     B 2018-01-06             NA NA
# 11       2     A 2018-01-07              2  2