Calculate value with two columns based on four conditions in R

Question

I have a large dataset uploaded in r (see below for a short version): I want to calculate a value for each Cruiseid, Samplenr, Species and Age (so based on four conditions):

Cruiseid    Samplenr Species Age Length LK  TNumStat    TNumLK
197502      37        154   0   12,5    2   2,791666667 5,583333
197502      37        154   0   17,5    3   2,166666667 6,5
197502      37        154   2   172,5   34  11,54166667 392,4167
197502      37        154   2   177,5   35  12,0625 422,1875
197502      37        154   2   182,5   36  2,083333333 75
197502      35        154   0   112,5   22  11,85654008 260,8439
197502      35        154   2   197,5   39  2,109704641 82,27848
197502      35        154   2   217,5   43  2,109704641 90,7173
197502      35        154   2   232,5   46  2,109704641 97,04641
197502      36        154   0   12,5    2   4,685314685 9,370629
197502      36        154   2   182,5   36  3,496503497 125,8741
197502      41        154   0   17,5    3   2,260869565 6,782609
197502      41        154   2   202,5   40  4,347826087 173,913
197502      41        154   2   212,5   42  2,173913043 91,30435
197502      41        154   2   242,5   48  2,173913043 104,3478
197503      56        154   0   17,5    3   7,428571429 22,28571
197503      56        154   0   147,5   29  10,30952381 298,9762
197503      56        154   2   172,5   34  13,19047619 448,4762
197503      56        154   2   187,5   37  2,380952381 88,09524
197503      54        154   0   12,5    2   3,35        6,7
197503      54        154   0   157,5   31  12          372
197503      54        154   0   167,5   33  13,25       437,25
197503      54        154   2   172,5   34  13,85       470,9
197503      54        154   2   187,5   37  2,5         92,5
197503      54        154   2   217,5   43  2,5         107,5
197503      53        154   0   12,5    2   2,875536481 5,751073
197503      53        154   0   97,5    19  4,806866953 91,33047
197503      53        154   0   107,5   21  5,622317597 118,0687
197503      53        154   0   142,5   28  8,776824034 245,7511

I want to calcuate:((TNumStat$TNumLK/TNumStat$TNumStat)*0.5+0.25)*10for each Cruiseid, Samplenr, Species and Age.

I have already tried something in a loop construction:

#######################
Cruise <- unique(TNumStat$Cruiseid)
Track <- unique(TNumStat$Samplenr)
#######################
AvrLengthCr <- c()
AvrLengthCr <- rep(NA, length(TNumStat$Species))
#######################
for(j in 1:length(Cruise)){
  t1.ss <- which(TNumStat$Cruiseid ==  Cruise[j])
  ###
  for(i in 1:length(Track)){
    t2.ss <- which(TNumStat$Samplenr[t1.ss] ==  Track[i])
    ###
    AvrLengthCr[t1.ss][t2.ss] <- ((TNumStat$TNumLK[t1.ss][t2.ss]/TNumStat$TNumStat[t1.ss][t2.ss])*0.5+0.25)*10
  }}

But it doesn't seem to work. And I've also been looking at something with dcast:

TNumStat2<-dcast(TNumStat,Cruiseid+Samplenr+Species+Age,formula = (((TNumStat$TNumLK/TNumStat$TNumStat*0.5+0.25)*10) )),na.rm=TRUE)

Non of the options I have tried seem to work, and I dont know how to solve this. Can someone please help me?

Thank you

Do you need `library(dplyr); df %>% group_by(Cruiseid, Samplenr, Species, Age) %>% mutate(ratio = ((TNumLK/TNumStat)*0.5+0.25)*10)` ? — Ronak Shah, Mar 13 '19 at 07:26
You probably have to clean your data fist with `TNumStat[c("TNumStat", "TNumLK")] <- lapply(TNumStat[c("TNumStat", "TNumLK")], function(x) as.numeric(gsub(",", ".", x)))` to get real decimal points and numeric values. — jay.sf, Mar 13 '19 at 07:39
@jay.sf: Why wont it work now than? Are the numbers not correct now? — Suusie, Mar 13 '19 at 08:02

score 1 · Accepted Answer · answered Mar 13 '19 at 07:26

Good Morning,

the question is not totally clear in my opinion. But you could try something like (with dplyr)

sample <- sample %>%
  mutate(calculate = ((TNumLK/TNumStat) * 0.5 + 0.25) * 10) %>%
  group_by(Cruiseid, Samplenr, Species, Age)

summarisedDF <- sample %>%
  summarise(avg.calculate = mean(calculate))

score 1 · Answer 2 · answered Mar 13 '19 at 08:19

What strikes me is your columns "Length", "TNumStat", "TNumLK" have , instead of . and thus are in character format that can not be coerced to numeric so easily.

TNumStat[c("TNumStat", "TNumLK")] <- 
  lapply(TNumStat[c("TNumStat", "TNumLK")], 
         function(x) as.numeric(gsub(",", ".", x)))

Maybe this is dependent to your system locale, so just ignore this step if it works for you.

Then, you could use by to apply your formula.

l <- by(TNumStat, TNumStat[c("Cruiseid", "Samplenr", "Species")],
        function(x) cbind(unique(x[1:3]),
                          value=with(x, ((mean(TNumLK)/mean(TNumStat))*0.5+0.25)*10)))

This gives you a list that you rbind to get the result.

TNumStat.new <- do.call(rbind, l)

TNumStat.new
#    Cruiseid Samplenr Species     value
# 6    197502       35     154 148.46288
# 10   197502       36     154  85.14956
# 1    197502       37     154 149.61421
# 12   197502       41     154 174.24600
# 26   197503       53     154 106.86347
# 20   197503       54     154 159.17545
# 16   197503       56     154 131.26698

Data

TNumStat <- structure(list(Cruiseid = c(197502L, 197502L, 197502L, 197502L, 
197502L, 197502L, 197502L, 197502L, 197502L, 197502L, 197502L, 
197502L, 197502L, 197502L, 197502L, 197503L, 197503L, 197503L, 
197503L, 197503L, 197503L, 197503L, 197503L, 197503L, 197503L, 
197503L, 197503L, 197503L, 197503L), Samplenr = c(37L, 37L, 37L, 
37L, 37L, 35L, 35L, 35L, 35L, 36L, 36L, 41L, 41L, 41L, 41L, 56L, 
56L, 56L, 56L, 54L, 54L, 54L, 54L, 54L, 54L, 53L, 53L, 53L, 53L
), Species = c(154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 
154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 
154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L, 154L), 
    Age = c(0L, 0L, 2L, 2L, 2L, 0L, 2L, 2L, 2L, 0L, 2L, 0L, 2L, 
    2L, 2L, 0L, 0L, 2L, 2L, 0L, 0L, 0L, 2L, 2L, 2L, 0L, 0L, 0L, 
    0L), Length = structure(c(3L, 8L, 9L, 10L, 11L, 2L, 13L, 
    16L, 17L, 3L, 11L, 8L, 14L, 15L, 18L, 8L, 5L, 9L, 12L, 3L, 
    6L, 7L, 9L, 12L, 16L, 3L, 19L, 1L, 4L), .Label = c("107,5", 
    "112,5", "12,5", "142,5", "147,5", "157,5", "167,5", "17,5", 
    "172,5", "177,5", "182,5", "187,5", "197,5", "202,5", "212,5", 
    "217,5", "232,5", "242,5", "97,5"), class = "factor"), LK = c(2L, 
    3L, 34L, 35L, 36L, 22L, 39L, 43L, 46L, 2L, 36L, 3L, 40L, 
    42L, 48L, 3L, 29L, 34L, 37L, 2L, 31L, 33L, 34L, 37L, 43L, 
    2L, 19L, 21L, 28L), TNumStat = structure(c(16L, 11L, 2L, 
    5L, 9L, 3L, 10L, 10L, 10L, 21L, 19L, 13L, 20L, 12L, 12L, 
    24L, 1L, 6L, 14L, 18L, 4L, 7L, 8L, 15L, 15L, 17L, 22L, 23L, 
    25L), .Label = c("10,30952381", "11,54166667", "11,85654008", 
    "12", "12,0625", "13,19047619", "13,25", "13,85", "2,083333333", 
    "2,109704641", "2,166666667", "2,173913043", "2,260869565", 
    "2,380952381", "2,5", "2,791666667", "2,875536481", "3,35", 
    "3,496503497", "4,347826087", "4,685314685", "4,806866953", 
    "5,622317597", "7,428571429", "8,776824034"), class = "factor"), 
    TNumLK = structure(c(16L, 18L, 11L, 12L, 21L, 8L, 22L, 25L, 
    29L, 24L, 4L, 20L, 5L, 26L, 1L, 6L, 9L, 14L, 23L, 19L, 10L, 
    13L, 15L, 28L, 2L, 17L, 27L, 3L, 7L), .Label = c("104,3478", 
    "107,5", "118,0687", "125,8741", "173,913", "22,28571", "245,7511", 
    "260,8439", "298,9762", "372", "392,4167", "422,1875", "437,25", 
    "448,4762", "470,9", "5,583333", "5,751073", "6,5", "6,7", 
    "6,782609", "75", "82,27848", "88,09524", "9,370629", "90,7173", 
    "91,30435", "91,33047", "92,5", "97,04641"), class = "factor")), class = "data.frame", row.names = c(NA, 
-29L))

Consider to accept as answer or vote up if you don't mind ;) — jay.sf, Mar 13 '19 at 09:13
I did, but I could press the 'up' button? :/...so I made the thingie green! — Suusie, Mar 13 '19 at 12:02
Ah, it's because you didn't have 15 reputations yet. Since you have them now it should work now. :) — jay.sf, Mar 13 '19 at 12:14

Calculate value with two columns based on four conditions in R

2 Answers2