R delete fathers row based on sons in hierarchycal data - r

Question

I'm working with some data like these:
id <- c(1,1,1,2,2,2,3,3,3,4,4) # fathers
name <- c('a','b','k','b','e','g','e','f','k','f','u') # sons
data <- data.frame(id,name)
data
> data
id name
1 1 a
2 1 b
3 1 k
4 2 b
5 2 e
6 2 g
7 3 e
8 3 f
9 3 k
10 4 f
11 4 u
My goal is this: if there is only a son that I do not want, remove all the row with the same father of the disliked son. For example, I don't like the son e, the result should be:
> data_e
id name
1 1 a
2 1 b
3 1 k
# 4 2 b
# 5 2 e
# 6 2 g
# 7 3 e
# 8 3 f
# 9 3 k
10 4 f
11 4 u
Because the rows with id 2 and 3 have in their name e.
This could be also a task like " I do not like e and f together":
> data_eandf
id name
1 1 a
2 1 b
3 1 k
4 2 b
5 2 e
6 2 g
# 7 3 e
# 8 3 f
# 9 3 k
10 4 f
11 4 u
Or, "I don't want you if you have e or f":
> data_eorf
id name
1 1 a
2 1 b
3 1 k
# 4 2 b
# 5 2 e
# 6 2 g
# 7 3 e
# 8 3 f
# 9 3 k
# 10 4 f
# 11 4 u
As you've noticed, to be more clear, I've "commented" the must-be-deleted rows.
I've searched, but I've found a lot of question based on only one column like data[which(data$name=='e'),], but this is going to remove only at sons' levels, not all the row of the relative father. Also I've thought to put the data in the wide format, paste all the name of a id in an unique cell, and fetch if there is e for example with function like grepl(), but I think this could be a problem with large dataset (these data are an example).
Do you have any idea about how to manage this?
Thanks in advance

Answer 1

Here's a function to handle the different cases
dislike1 <- c('e')
dislike2 <- c('e', 'f')
myfun <- function(df, dislike, ops = NULL) {
require(dplyr)
if (is.null(ops) || ops == 'OR') {
df %>%
group_by(id) %>%
filter(!any(name %in% dislike)) %>%
ungroup
} else if (ops == 'AND') {
df %>%
group_by(id) %>%
filter(!all(dislike %in% name)) %>%
ungroup
}
}
myfun(data, dislike1)
# A tibble: 5 x 2
# id name
# <dbl> <fct>
# 1 1 a
# 2 1 b
# 3 1 k
# 4 4 f
# 5 4 u
myfun(data, dislike2, 'AND')
# A tibble: 8 x 2
# id name
# <dbl> <fct>
# 1 1 a
# 2 1 b
# 3 1 k
# 4 2 b
# 5 2 e
# 6 2 g
# 7 4 f
# 8 4 u
myfun(data, dislike2, 'OR')
# A tibble: 3 x 2
# id name
# <dbl> <fct>
# 1 1 a
# 2 1 b
# 3 1 k

Answer 2

data[!(data$id %in% unique(data[data$name == 'e', 'id'])),]
unique(data[data$name == 'e', 'id']) will get the unique id's that have 'e' in the name field. Then you can use the %in% operator to find all the rows with those id's. The ! is a negation operator.

Answer 3

I have a data.table solution
require(data.table)
id <- c(1,1,1,2,2,2,3,3,3,4,4) # fathers
name <- c('a','b','k','b','e','g','e','f','k','f','u') # sons
data <- data.table(id,name)
# names to be deleted
to_del <- c("e","f")
# returns only id's without any of the names to be deleted
data[ , .SD[ !any(name %in% to_del) ,name ] , by = "id"]
id V1
1: 1 a
2: 1 b
3: 1 k