Search code examples
rcut

Manipulation on levels produced by cut R function


I want to do some manipulation on levels produced by cut R function. I want to have exp(Labs) in my MWE.

set.seed(12345)
Y <- rnorm(n = 50, mean = 500, sd = 1)
Y1 <-  cut(log(Y), 5)
Labs <- levels(Y1)
Labs
[1] "(6.21,6.212]"  "(6.212,6.213]" "(6.213,6.215]" "(6.215,6.217]" "(6.217,6.219]"

exp(cbind(lower = as.numeric( sub("\\((.+),.*", "\\1", Labs) ),
      upper = as.numeric( sub("[^,]*,([^]]*)\\]", "\\1", Labs) )))
        lower    upper

[1,] 497.7013 498.6976
[2,] 498.6976 499.1966
[3,] 499.1966 500.1960
[4,] 500.1960 501.1974
[5,] 501.1974 502.2008

Question

How I can get exp(Labs) here?

Desired Output

"(497.7013, 498.6976]"  "(498.6976, 499.1966]" "(499.1966, 500.1960]" "(500.1960, 501.1974]"  "(501.1974, 502.2008]" 

Edited

Based on @akrun answer:

    Labs1 <- c("(-2.32,0.99]", "(0.99,4.28]", "(4.28,7.58]", "(7.58,10.9]", "(10.9,14.2]")

Labs1
    [1] "(-2.32,0.99]" "(0.99,4.28]"  "(4.28,7.58]"  "(7.58,10.9]"  "(10.9,14.2]"

    exp(cbind(lower = as.numeric( sub("\\((.+),.*", "\\1", Labs1) ),
              upper = as.numeric( sub("[^,]*,([^]]*)\\]", "\\1", Labs1) )))

            lower        upper
[1,] 9.827359e-02 2.691234e+00
[2,] 2.691234e+00 7.224044e+01
[3,] 7.224044e+01 1.958629e+03
[4,] 1.958629e+03 5.417636e+04
[5,] 5.417636e+04 1.468864e+06


    gsubfn('([0-9.]+)', ~round(exp(as.numeric(x)),4), Labs1)

[1] "(-10.1757,2.6912]"         "(2.6912,72.2404]"          "(72.2404,1958.629]"       
[4] "(1958.629,54176.3638]"     "(54176.3638,1468864.1897]"

    res1 <-  exp(as.data.frame(t(sapply(strsplit(Labs1, '[^0-9.]+'), 
                                       function(x) as.numeric(x[-1])))))
    sprintf('(%s]', do.call(paste, c(round(res1,4), sep=", ")))


[1] "(10.1757, 2.6912]"          "(2.6912, 72.2404]"          "(72.2404, 1958.629]"       
[4] "(1958.629, 54176.3638]"     "(54176.3638, 1468864.1897]"

Solution

  • A compact option would be to use gsubfn. We match the numeric elements with dots (([0-9.]+)) in the pattern argument, and we replace the matched one by first converting it to 'numeric', take the exp and round.

    library(gsubfn)
    gsubfn('([-0-9.]+)', ~round(exp(as.numeric(x)),4), Labs)
    #[1] "(497.7013,498.6976]" "(498.6976,499.1966]" "(499.1966,500.196]" 
    #[4] "(500.196,501.1974]"  "(501.1974,502.2008]"
    

    NOTE: This depends on the pattern we use.


    Or another option to avoid two calls to sub would be strsplit. We split on the non-numeric elements. The output will be a list so we can use either lapply/sapply to loop over the list elements, convert to numeric class, and create a 'data.frame' with two columns.

    res <-  exp(as.data.frame(t(sapply(strsplit(Labs, '[^-0-9.]+'), 
                  function(x) as.numeric(x[-1])))))
    

    Regarding getting the expected output based on the OP's code. I changed cbind to data.frame in the original code so that do.call can be used.

     res <-  exp(data.frame(lower = as.numeric( sub("\\((.+),.*", "\\1", Labs) ),
             upper = as.numeric( sub("[^,]*,([^]]*)\\]", "\\1", Labs) )))
    

    We paste the row elements of 'res' (do.call(paste0) and then add the additional parentheses with either sprintf or another paste

      sprintf('(%s]', do.call(paste, c(round(res,4), sep=", ")))
      #[1] "(497.7013, 498.6976]" "(498.6976, 499.1966]" "(499.1966, 500.196]" 
      #[4] "(500.196, 501.1974]"  "(501.1974, 502.2008]"
    

    Update

    Checking the output with the 'Labs1'

     gsubfn('([-0-9.]+)', ~round(exp(as.numeric(x)),4), Labs1)
     #[1] "(0.0983,2.6912]"           "(2.6912,72.2404]"         
     #[3] "(72.2404,1958.629]"        "(1958.629,54176.3638]"    
     #[5] "(54176.3638,1468864.1897]"
    
    
     exp(as.data.frame(t(sapply(strsplit(Labs1, '[^-0-9.]+'), 
                 function(x) as.numeric(x[-1])))))
     #           V1           V2
     #1 9.827359e-02 2.691234e+00
     #2 2.691234e+00 7.224044e+01
     #3 7.224044e+01 1.958629e+03
     #4 1.958629e+03 5.417636e+04
     #5 5.417636e+04 1.468864e+06