find last row with to multiple groups - r

Question

| id | msgid | source | value |
|----|-------|--------|-------|
| 1 | 1 | B | 0 |
| 1 | 2 | A | 1 |
| 1 | 3 | B | 0 |
| 2 | 1 | B | 0 |
| 2 | 2 | A | 0 |
| 2 | 3 | A | 1 |
| 2 | 4 | B | 0 |
In the above snippet, I want to create column value from the other columns. id is a conversation and msgId is the message in each conversation.
I wish to identify the row number for the last message that came from source=A.
I made an attempt to solve it. However, I was able to identify only the last row within a conversation.
last_values <- dat %>% group_by(id) %>%
slice(which.max(msgid)) %>%
ungroup %>%
mutate(value = cumsum(msgid))
dat$final_val <- 0
dat[last_values$value,5] <- 1

Answer 1

We can create the column 'value' by
dat %>%
group_by(id) %>%
mutate(value1 = as.integer(source == "A" & !duplicated(source == "A", fromLast = TRUE)))
# A tibble: 7 x 5
# Groups: id [2]
# id msgid source value value1
# <int> <int> <chr> <int> <int>
#1 1 1 B 0 0
#2 1 2 A 1 1
#3 1 3 B 0 0
#4 2 1 B 0 0
#5 2 2 A 0 0
#6 2 3 A 1 1
#7 2 4 B 0 0

Answer 2

Another dplyr solution:
library(dplyr)
# create data
df <- data.frame(
id = c(1, 1, 1, 2, 2, 2, 2),
msgid = c(1, 2, 3, 1, 2, 3, 4),
source = c("B", "A", "B", "B", "A", "A", "B")
)
df <- df %>%
group_by(id, source) %>% # group by id and source
mutate(value = as.integer(ifelse((row_number() == n()) & source == "A", 1, 0))) # write 1 if it's the last occurence of a group and the source is "A"
> df
# A tibble: 7 x 4
# Groups: id, source [4]
id msgid source value
<dbl> <dbl> <fctr> <dbl>
1 1 1 B 0
2 1 2 A 1
3 1 3 B 0
4 2 1 B 0
5 2 2 A 0
6 2 3 A 1
7 2 4 B 0

Answer 3

I came up with the following solution
library(tidyverse)
# first we create the dataframe as it wasn't supplied in the question
df <- tibble(
id = c(1, 1, 1, 2, 2, 2, 2),
msgid = c(1, 2, 3, 1, 2, 3, 4),
source = c("B", "A", "B", "B", "A", "A", "B")
)
df %>%
# group by both id and source
group_by(id, source) %>%
mutate(
# create a new column
value = max(msgid) == msgid & source == "A",
# convert the new column to integers
value = as.integer(value)
)
Output:
# A tibble: 7 x 4
# Groups: id, source [4]
id msgid source value
<dbl> <dbl> <chr> <int>
1 1 1 B 0
2 1 2 A 1
3 1 3 B 0
4 2 1 B 0
5 2 2 A 0
6 2 3 A 1
7 2 4 B 0

Answer 4

I used index flagging for finding the final position of A and checked if that number matches with row number in order to assign 1 to value.
library(dplyr)
mydf <- data.frame(id = c(1, 1, 1, 2, 2, 2, 2),
msgid = c(1, 2, 3, 1, 2, 3, 4),
source = c("B", "A", "B", "B", "A", "A", "B"))
group_by(mydf, id) %>%
mutate(value = if_else(last(grep(source, pattern = "A")) == row_number(),
1, 0)
id msgid source value
<dbl> <dbl> <fctr> <dbl>
1 1.00 1.00 B 0
2 1.00 2.00 A 1.00
3 1.00 3.00 B 0
4 2.00 1.00 B 0
5 2.00 2.00 A 0
6 2.00 3.00 A 1.00
7 2.00 4.00 B 0