在r的数据框中找到每行的第一，第二和第三最大值及其对应的列名

Question

I am trying to find first max, second max, and third max value and corresponding col names for each row, but unable to do that in r. Please help.

Here is how the dataframe looks like:

              X1    X2    X3   X4    X5   X6   X7    X8    X9    X10   X11  X12   
      10003   0.0   0.0   0.0  0.0   0.0  0.0  0.0   0.0   0.0   0.0   0.0  0.0       
      10006   0.0   0.0   0.0  0.0   0.0  0.0 16.7   0.0   0.0   0.0   0.0  0.0       
      10007   0.0   0.0   0.0  0.0   0.0  0.0  0.0   0.0   0.0   0.0   0.0  0.0       
      10008   0.0   0.0   0.0  0.0   0.0  0.0  0.0   0.0   0.0   0.0   0.0  0.0       
      10010   0.0   0.0   0.0  0.0   0.0  0.0  0.0   0.0   0.0   0.0   0.0  0.0       
      10014   0.0   0.0   0.0  0.0   0.0  0.0  0.0   0.0   0.0   0.0   0.0  0.0   

Answer 1

This is the sample data you posted in your comment:

data <-read.table(text="       x1    x2    x3     x4    x5    x6   x7   x8    x9
                        1003    0  45.7     0   22.9     0  13.7    0    0  23.1 
                        1004 22.2     0  13.2      0   5.4     0  9.7    0     0 
                        1005    0     0     0     12   2.1     0    0  3.2     0  
                        1006  1.2     0   1.2      0  43.9  43.9    0    0  57.6",
                    header=T)

You can use dplyr and tidyverse to acheive this.

---

The following code will give you the maximum three columns across all the rows:

library(dplyr)
library(tidyverse)

data %>% 
  rownames_to_column() %>%
  gather(column, value, -rowname) %>%
  group_by(rowname) %>% 
  arrange(desc(value)) %>% 
  head(3) 

This will give you the following result:

# A tibble: 3 x 3
# Groups:   rowname [3]
#   rowname column value
#   <chr>   <chr>  <dbl>
# 1 1006    x9      57.6
# 2 1003    x2      45.7
# 3 1006    x5      43.9

---

If you want to get the maximum three values for each row, you can do it as follows:

result <- data %>% 
  rownames_to_column() %>%
  gather(column, value, -rowname) %>%
  group_by(rowname) %>% 
  mutate(max = rank(-value)) %>%
  filter(max <= 3) %>% 
  arrange(rowname, max)

Which will give you the following result:

# A tibble: 12 x 4
# Groups:   rowname [4]
#    rowname column value   max
#    <chr>   <chr>  <dbl> <dbl>
#  1 1003    x2      45.7   1  
#  2 1003    x9      23.1   2  
#  3 1003    x4      22.9   3  
#  4 1004    x1      22.2   1  
#  5 1004    x3      13.2   2  
#  6 1004    x7       9.7   3  
#  7 1005    x4      12     1  
#  8 1005    x8       3.2   2  
#  9 1005    x5       2.1   3  
# 10 1006    x9      57.6   1  
# 11 1006    x5      43.9   2.5
# 12 1006    x6      43.9   2.5

---

To summarize the result for each row, use the following code:

result %>% 
  mutate(result = paste0(column, "=", value, collapse = ", ")) %>% 
  select(result) %>% 
  distinct()

Which will give you the following result:

# A tibble: 4 x 2
# Groups:   rowname [4]
#   rowname result                   
#   <chr>   <chr>                    
# 1 1003    x2=45.7, x9=23.1, x4=22.9
# 2 1004    x1=22.2, x3=13.2, x7=9.7 
# 3 1005    x4=12, x8=3.2, x5=2.1    
# 4 1006    x9=57.6, x5=43.9, x6=43.9

Hope it helps.

Answer 2

Here is my approach:

 # Make up data because yours is pretty unreproducible:
 df <- data.frame(X1=1:5, X2=c(3,5,1,6,7))

 # combine and sort the data by decreasing value:
 a <- sort(dplyr::combine(df), decreasing = T)[1:3]

 # For loop to get the indexes:
 for(i in 1:length(a)){
    print(which(df==a[i], arr.ind = T))
 }

This will give you what you need. Replace print with whatever you want to do (eg assign or whatever you need)

Answer 3

You can use

max.names = apply(data, 1, function(x) names(sort(x, decreasing = T)[1:3]))
max.vals = apply(data, 1, function(x) sort(x, decreasing = T)[1:3])
data = cbind(data, t(max.names), t(max.vals))
#        x1   x2   x3   x4   x5   x6  x7  x8   x9  1  2  3    1    2    3
# 1003  0.0 45.7  0.0 22.9  0.0 13.7 0.0 0.0 23.1 x2 x9 x4 45.7 23.1 22.9
# 1004 22.2  0.0 13.2  0.0  5.4  0.0 9.7 0.0  0.0 x1 x3 x7 22.2 13.2  9.7
# 1005  0.0  0.0  0.0 12.0  2.1  0.0 0.0 3.2  0.0 x4 x8 x5 12.0  3.2  2.1
# 1006  1.2  0.0  1.2  0.0 43.9 43.9 0.0 0.0 57.6 x9 x5 x6 57.6 43.9 43.9