问题
I want to reshape/ rearrange a dataset, that is stored as a data.frame with 2 columns:
- id (non-unique, i.e. can repeat over several rows) --> stored as character
- value --> stored as numeric value (range 1:3)
Sample data:
id <- as.character(1001:1003)
val_list <- data.frame(sample(1:3, size=12, replace=TRUE))
have <- data.frame(cbind(rep(id, 4), val_list))
colnames(have) <- c("id", "values")
have <- have %>% arrange(id)
This gives me the following output:
id values
1 1001 2
2 1001 2
3 1001 2
4 1001 3
5 1002 2
6 1002 3
7 1002 2
8 1002 2
9 1003 1
10 1003 3
11 1003 1
12 1003 2
What I want:
want <- data.frame(cbind(have[1:4, 2],
have[5:8, 2],
have[9:12, 2]))
colnames(want) <- id
Output of want:
1001 1002 1003
1 2 2 1
2 2 3 3
3 2 2 1
4 3 2 2
My original dataset has >1000 variables "id" and >50 variables "value". I want to chunk/ slice the dataset get a new data.frame where each "id" variable will represent one column listing its "value" variable content.
It is possible to solve it via a loop, but I want to have the vectorized solution. If possible with base R as "one-liner", but other solutions also appreciated.
回答1:
You can create a unique row value for each id
and use pivot_wider
.
have %>%
group_by(id) %>%
mutate(row = row_number()) %>%
tidyr::pivot_wider(names_from = id, values_from = values) %>%
select(-row)
# A tibble: 4 x 3
# `1001` `1002` `1003`
# <int> <int> <int>
#1 1 3 1
#2 3 2 3
#3 2 2 3
#4 2 2 3
Or using data.table
library(data.table)
dcast(setDT(have), rowid(id)~id, value.var = 'values')
data
df <- structure(list(id = c(1001L, 1001L, 1001L, 1001L, 1002L, 1002L,
1002L, 1002L, 1003L, 1003L, 1003L, 1003L), values = c(2L, 2L,
2L, 3L, 2L, 3L, 2L, 2L, 1L, 3L, 1L, 2L)), class = "data.frame",
row.names = c(NA, -12L))
来源:https://stackoverflow.com/questions/61867904/reshaping-data-frame-with-a-by-group-where-id-variable-repeats