Gather or transpose data with multiple rows as 'key' argument - r

Question

In my mind I want to tidyr::gather() gather on not only the column names but also on row 1 and 2. What I want to achieve is to have a data frame with 5 columns and 4 rows.
This is a little piece of the dataset I'm working with:
library(tidyverse)
# A tibble: 4 x 3
Aanduiding `Coolsingel 40 links` `Goudseweg 15 links`
<chr> <chr> <chr>
1 Gebiedsnummer 1 2
2 Postcode 3011 AD 3031 XH
3 Leefbaar Rotterdam 124 110
4 Partij van de Arbeid (P.v.d.A.) 58 65
and its reproducable dput(df) to work with:
df <- structure(list(Aanduiding = c("Gebiedsnummer", "Postcode", "Leefbaar Rotterdam",
"Partij van de Arbeid (P.v.d.A.)"), `Coolsingel 40 links` = c("1",
"3011 AD", "124", "58"), `Goudseweg 15 links` = c("2", "3031 XH",
"110", "65")), row.names = c(NA, -4L), class = c("tbl_df", "tbl",
"data.frame"), .Names = c("Aanduiding", "Coolsingel 40 links",
"Goudseweg 15 links"))
So wanted out put looks like this:
Aanduiding Gebiedsnummer Postcode adres value
<chr> <dbl> <chr> <chr> <dbl>
1 Leefbaar Rotterdam 1.00 3011 AD Coolsingel 40 links 124
2 Leefbaar Rotterdam 1.00 3031 XH Goudseweg 15 links 120
3 Partij van de Arbeid (P.v.d.A.) 2.00 3011 AD Coolsingel 40 links 58.0
4 Partij van de Arbeid (P.v.d.A.) 2.00 3031 XH Goudseweg 15 links 65.0
I use the gather() function from the tidyr package a lot, but this is alway when I only want to gather the column names with a certain value. Now I actually want to gather the column names but also observation on row 1 and 2.
Can I gather on multiple key's? Or paste the values in observation 1 and 2 to the column, then gather() and then separate()?
What's the best tactic here, if possible in a tidyr way.
Much appreciated.

Answer 1

There's two things that need to be done here, and you'll have to figure out how to break down your dataset accordingly.
data.frame(t(df[1:2,]))
will give you:
X1 X2
Aanduiding Gebiedsnummer Postcode
Coolsingel 40 links 1 3011 AD
Goudseweg 15 links 2 3031 XH
And
tidyr::gather(df[3:4,],key="adres",value="value", `Coolsingel 40 links`, `Goudseweg 15 links`)
will give you:
Aanduiding adres value
<chr> <chr> <chr>
1 Leefbaar Rotterdam Coolsingel 40 links 124
2 Partij van de Arbeid (P.v.d.A.) Coolsingel 40 links 58
3 Leefbaar Rotterdam Goudseweg 15 links 110
4 Partij van de Arbeid (P.v.d.A.) Goudseweg 15 links 65
How you go on from there is another problem, possibly a left_join based on adres, but that really depends on how the rest of the data is structured.

Answer 2

You can do this with a combination of gather and spread a few times. I do this often when I need to move a value out to serve as a denominator for a calculation.
library(tidyverse)
...
The goal is to move Gebiedsnummer and Postcode out of Aanduiding, and to gather the other two columns into one column of values. The first gather gets you this:
df %>%
gather(key = address, value = value, -Aanduiding)
#> # A tibble: 8 x 3
#> Aanduiding address value
#> <chr> <chr> <chr>
#> 1 Gebiedsnummer Coolsingel 40 links 1
#> 2 Postcode Coolsingel 40 links 3011 AD
#> 3 Leefbaar Rotterdam Coolsingel 40 links 124
#> 4 Partij van de Arbeid (P.v.d.A.) Coolsingel 40 links 58
#> 5 Gebiedsnummer Goudseweg 15 links 2
#> 6 Postcode Goudseweg 15 links 3031 XH
#> 7 Leefbaar Rotterdam Goudseweg 15 links 110
#> 8 Partij van de Arbeid (P.v.d.A.) Goudseweg 15 links 65
Using a spread after that gets:
df %>%
gather(key = address, value = value, -Aanduiding) %>%
spread(key = Aanduiding, value = value)
#> # A tibble: 2 x 5
#> address Gebiedsnummer `Leefbaar Rotter… `Partij van de Arbe… Postcode
#> <chr> <chr> <chr> <chr> <chr>
#> 1 Coolsinge… 1 124 58 3011 AD
#> 2 Goudseweg… 2 110 65 3031 XH
Then you want to gather again, but to keep address, Gebiedsnummer, and Postcode as their own columns. The select is just there to get the columns in order. So all together:
df %>%
gather(key = address, value = value, -Aanduiding) %>%
spread(key = Aanduiding, value = value) %>%
gather(key = Aanduiding, value = value, -Gebiedsnummer, -address, -Postcode) %>%
select(Aanduiding, Gebiedsnummer, Postcode, address, value) %>%
mutate_at(vars(Gebiedsnummer, value), as.numeric)
#> # A tibble: 4 x 5
#> Aanduiding Gebiedsnummer Postcode address value
#> <chr> <dbl> <chr> <chr> <dbl>
#> 1 Leefbaar Rotterdam 1 3011 AD Coolsingel 40 l… 124
#> 2 Leefbaar Rotterdam 2 3031 XH Goudseweg 15 li… 110
#> 3 Partij van de Arbeid (P.v… 1 3011 AD Coolsingel 40 l… 58
#> 4 Partij van de Arbeid (P.v… 2 3031 XH Goudseweg 15 li… 65
Created on 2018-08-24 by the reprex package (v0.2.0).