R select values for each group - r

Question

For each combination of WEEK and GROUP I wish to find the day (1 or 2) when SCORE is lowest for TYPE == 'M' and then keep the TYPE == 'M' and TYPE == 'E' value on that same day.
HAVE = data.frame("WEEK" = c(1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4),
"GROUP" = c('A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B'),
"SCORE" = c(29, 10, 19, 11, 28, 38, 23, 27, 32, 19, 32, 26, 35, 27, 39, 51, 43, 12, 35, 29, 14, 47, 12, 36, 17, 11, 24, 19, 19, 12, 33, 7),
"DAY" = c(1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2, 1, 1, 2, 2),
"TYPE" = c('M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E'))
WANT=data.frame("WEEK"=c(1, 1, 2, 2, 3, 3, 4, 4, 1, 1, 2, 2, 3, 3, 4, 4),
"GROUP"=c('A', 'A', 'A', 'A', 'A', 'A', 'A', 'A', 'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B'),
"SCORE"=c(19, 11, 23, 27, 32, 19, 35, 27, 35, 29, 12, 36, 17, 11, 19, 12),
"DAY"=c(2, 2, 2, 2, 1, 1, 1, 1, 2, 2, 2, 2, 1, 1, 1, 1),
"TYPE"=c('M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E', 'M', 'E'),
"SCORE.MIN"=c(1, NA, 1, NA, 1, NA, 1, NA, 1, NA, 1, NA, 1, NA, 1, NA))
I have 'HAVE' and wish to create 'WANT'
For each combination of WEEK and GROUP, select the DAY when TYPE = M and SCORE is lowest. This I am able to do by
WANT.ATTEMPT = HAVE %>%
group_by(WEEK, GROUP) %>%
mutate(SCORE.MIN = ifelse(TYPE == 'M', min(SCORE), NA))
HOWEVER what I wish for is the dataset WANT where as you see I group by WEEK and GROUP and find the DAY when SCORE is minimum for TYPE == 'M' but I also keep the SCORE for TYPE == 'E' for the same day

Answer 1

Update to prevent ties.
You can do:
HAVE %>%
group_by(WEEK, GROUP) %>%
filter(DAY == first(DAY[SCORE == min(SCORE[TYPE == 'M'])]) | TYPE == 'E') %>%
filter(DAY %in% DAY[TYPE == 'M']) %>%
ungroup()
# A tibble: 16 x 5
WEEK GROUP SCORE DAY TYPE
<dbl> <chr> <dbl> <dbl> <chr>
1 1 A 19 2 M
2 1 A 11 2 E
3 2 A 23 2 M
4 2 A 27 2 E
5 3 A 32 1 M
6 3 A 19 1 E
7 4 A 35 1 M
8 4 A 27 1 E
9 1 B 35 2 M
10 1 B 29 2 E
11 2 B 12 2 M
12 2 B 36 2 E
13 3 B 17 1 M
14 3 B 11 1 E
15 4 B 19 1 M
16 4 B 12 1 E

Answer 2

We could also arrange first before grouping and then do the filter
library(dplyr)
HAVE %>%
arrange(WEEK, GROUP, TYPE != 'M', SCORE) %>%
group_by(WEEK, GROUP) %>%
filter(DAY %in% DAY[TYPE == "M"][1]) %>%
ungroup
-output
# A tibble: 16 × 5
WEEK GROUP SCORE DAY TYPE
<dbl> <chr> <dbl> <dbl> <chr>
1 1 A 19 2 M
2 1 A 11 2 E
3 1 B 35 2 M
4 1 B 29 2 E
5 2 A 23 2 M
6 2 A 27 2 E
7 2 B 12 2 M
8 2 B 36 2 E
9 3 A 32 1 M
10 3 A 19 1 E
11 3 B 17 1 M
12 3 B 11 1 E
13 4 A 35 1 M
14 4 A 27 1 E
15 4 B 19 1 M
16 4 B 12 1 E