CODEWITHSUNDEEP
rdataframe
Patrick Parts
Patrick Parts

Reputation: 203

ordering dataframe using three variable at a same time

I would like to order a table using different variable as following:

Below is an example of what I mean. If not clear enough don't hesitate to ask more detail In the example line 5 (19,5,10) is before line 6 (6, NA, NA) because 5 is lower than 6

set.seed(123) 
data=data.frame(col1=sample(1:20,10), col2=c(sample(1:20,5), NA, NA, NA, NA, NA), col3=c(sample(1:20,5), NA, NA, NA, NA, NA))

  col1 col2 col3
1    15   14    7
2    19    5   10
3    14    9    9
4     3    3    4
5    10    8   14
6     2   NA   NA
7     6   NA   NA
8    11   NA   NA
9     5   NA   NA
10    4   NA   NA

> data_output
   col1 col2 col3
1     2   NA   NA
2     3    3    4
3     4   NA   NA
4     5   NA   NA
5    19    5   10
6     6   NA   NA
7    15   14    7
8    10    8   14
9    14    9    9
10   11   NA   NA

Any idea ?

thank you

Upvotes: 2

Views: 36

Answers (2)

harre
harre

Reputation: 7287

The core idea would be to arrange by the minimum value per row. Here using dplyr:

library(dplyr)

data |>
  rowwise() |>
  mutate(min_sort = min(c_across(everything()), na.rm = TRUE)) |>
  ungroup() |>
  arrange(min_sort) # |>
  #select(-min_sort)

Output:

# A tibble: 10 × 4
    col1  col2  col3 min_sort
   <int> <int> <int>    <int>
 1     2    NA    NA        2
 2     3     3     4        3
 3     4    NA    NA        4
 4    19     5    10        5
 5     5    NA    NA        5
 6     6    NA    NA        6
 7    15    14     7        7
 8    10     8    14        8
 9    14     9     9        9
10    11    NA    NA       11

Update: @Maël beat me with 18 sec, but now you have the same core idea expressed in different ways.

Upvotes: 2

Ma&#235;l
Ma&#235;l

Reputation: 52209

Take the minimum value per row, and order it:

data[order(apply(data, 1, min, na.rm = T)), ]

or, with pmin:

data[order(do.call(pmin, c(data, na.rm = TRUE))), ]

or with matrixStats::rowMins:

library(matrixStats)
data[order(rowMins(as.matrix(data), na.rm = T)), ]

   col1 col2 col3
6     2   NA   NA
4     3    3    4
10    4   NA   NA
2    19    5   10
9     5   NA   NA
7     6   NA   NA
1    15   14    7
5    10    8   14
3    14    9    9
8    11   NA   NA

Upvotes: 3

Related Questions