gather() with two key columns - r

Question

I have a dataset that has two rows of data, and want to tidy them using something like gather() but don't know how to mark both as key columns.
The data looks like:
Country US Canada US
org_id 332 778 920
02-15-20 25 35 54
03-15-20 30 10 60
And I want it to look like
country org_id date purchase_price
US 332 02-15-20 25
Canada 778 02-15-20 35
US 920 02-15-20 54
US 332 03-15-20 30
Canada 778 03-15-20 10
US 920 03-15-20 60
I know gather() can move the country row to a column, for example, but is there a way to move both the country and org_id rows to columns?

Answer 1

It is not a good idea to have duplicate column names in the data so I'll rename one of them.
names(df)[4] <- 'US_1'
gather has been retired and replaced with pivot_longer.
This is not a traditional reshape because the data in the 1st row needs to be treated differently than rest of the rows so we can perform the reshaping separately and combine the result to get one final dataframe.
library(dplyr)
library(tidyr)
df1 <- df %>% slice(-1L) %>% pivot_longer(cols = -Country)
df %>%
slice(1L) %>%
pivot_longer(-Country, values_to = 'org_id') %>%
select(-Country) %>%
inner_join(df1, by = 'name') %>%
rename(Country = name, date = Country) -> result
result
# Country org_id date value
# <chr> <int> <chr> <int>
#1 US 332 02-15-20 25
#2 US 332 03-15-20 30
#3 Canada 778 02-15-20 35
#4 Canada 778 03-15-20 10
#5 US_1 920 02-15-20 54
#6 US_1 920 03-15-20 60
data
df <- structure(list(Country = c("org_id", "02-15-20", "03-15-20"),
US = c(332L, 25L, 30L), Canada = c(778L, 35L, 10L), US = c(920L,
54L, 60L)), class = "data.frame", row.names = c(NA, -3L))

Answer 2

First, we paste together Country and org_id
library(tidyverse)
data <- set_names(data, paste(names(data), data[1,], sep = "-"))
data
Country-org_id US-332 Canada-778 US-920
1 org_id 332 778 920
2 02-15-20 25 35 54
3 03-15-20 30 10 60
Then, we drop the first row, pivot the table and separate the column name.
df <- data %>%
slice(2:n()) %>%
rename(date = `Country-org_id`) %>%
pivot_longer(cols = -date, values_to = "price") %>%
separate(col = name, into = c("country", "org_id"), sep = "-")
df
# A tibble: 6 x 4
date country org_id price
<chr> <chr> <chr> <int>
1 02-15-20 US 332 25
2 02-15-20 Canada 778 35
3 02-15-20 US 920 54
4 03-15-20 US 332 30
5 03-15-20 Canada 778 10
6 03-15-20 US 920 60