CODEWITHSUNDEEP
rdataframestring-comparison
Bernd V.
Bernd V.

Reputation: 193

Compare character in column with string in another column

Since some days I try to find a way to subset my data frame by comparing a character in a column with a string in another column.

In case the character is not within the string, I want to copy a value to a new column. I searched high and low, tried many examples, but for some reason I do not get it to work on my data frame.

    df <- structure(list(POLY = c("K3", "K3", "K3", "K4", "K4", "K4", "K4", 
    "K6", "K6", "K7", "K7", "K7", "L1", "L1", "L1"), FIX = c("O", 
    "K", "M", "M", "K", "O", "L", "K", "M", "K", "O", "M", "M", "L", 
    "O"), SESSTIME = c(310, 190, 181, 188, 151, 260, 268, 200, 259, 
    245, 180, 188, 259, 199, 244), CODE = c("KO", "KO", "KO", "KM", 
    "KM", "KM", "KM", "KM", "KM", "KO", "KO", "KO", "LMO", "LMO", 
    "LMO")), .Names = c("POLY", "FIX", "SESSTIME", "CODE"), row.names = c(42L, 
    44L, 46L, 115L, 116L, 117L, 133L, 225L, 231L, 269L, 270L, 328L, 
    420L, 425L, 431L), class = "data.frame")

This it what a part of it looks like:

    row.names   POLY    FIX SESSTIME    CODE    SESSTIME2
1   42          K3      O   310         KO      NA
2   44          K3      K   190         KO      NA
3   46          K3      M   181         KO      ...
4   115         K4      M   188         KM
5   116         K4      K   151         KM
6   117         K4      O   260         KM      NA
7   133         K4      L   268         KM      268
8   225         K6      K   200         KM      NA
9   231         K6      M   259         KM
10  269         K7      K   245         KO
11  270         K7      O   180         KO
12  328         K7      M   188         KO      188
13  420         L1      M   259        LMO
14  425         L1      L   199        LMO
15  431         L1      O   244        LMO

So when FIX is not in CODE the value of SESSTIME should be copied to SESSTIME2 (column already prepopulated with NA)

I tried it for example with 

  df$FIX %in% strsplit(as.character(df$CODE,""))

or similar, but the comparison is always TRUE.

All examples I found only applied (and worked) with comparison of a single character e.g. "K" hardcoded with a vector c("K","L","M") or so, but never an example how to apply this to data frame columns and rows.

I'm getting a little bit nervous ...

Anyone an idea what I'm doing wrong?

UPDATE:

Thanx to the answer below, my code now looks like this and does what I need:

df3$SESSTIME2[!(mapply(function(i, j) length(grep(i, j)), df$FIX, df$CODE)) & is.na(df$SESSTIME2)] 

<- 

df$SESSTIME[!(mapply(function(i, j) length(grep(i, j)), df$FIX, df$CODE)) & is.na(df$SESSTIME2)]

Upvotes: 0

Views: 2497

Answers (1)

csgillespie
csgillespie

Reputation: 60452

The reason your code doesn't work is because

strsplit(as.character(df$CODE,""))

returns a list. Instead, you need to use mapply to detect if there is a match.

Here we used grep which allows more flexible character matching

# The values of FIX & CODE are passed to i and j
mapply(function(i, j) length(grep(i, j)), df$FIX, df$CODE)

or using %in% 

## Suggested by akrun
mapply('%in%', df$FIX,strsplit(as.character(df$CODE), ''))

Upvotes: 2

Related Questions