How to replace values in dataframe based on a second dataframe in R? - r

Question

I have a dataframe df1 with multiple columns, each column representing a species name (sp1, sp2, sp3, ...).
df1
sp1 sp2 sp3 sp4
NA NA r1 r1
NA NA 1 3
NA 5 NA NA
m4 NA NA m2
I would like to replace each value in df1 with values based on a second dataframe, df2. Here, the values in df1 should match df2$scale_nr, and replaced by df2$percentage. Thus, the result should be so that I have my values in df1 based on $percentage in df2.
df2
scale_nr percentage
r1 1
p1 1
a1 1
m1 1
r2 2
p2 2
a2 2
m2 2
1 10
2 20
3 30
4 40
...
Then after replacement df1 should look like
df1
sp1 sp2 sp3 sp4
NA NA 1 1
NA NA 10 30
NA 50 NA NA
4 NA NA 2
I tried this:
df2$percentage[match(df1$sp1, df2$scale_nr)] # this one works for one column
which works for one column, I know I should be able to do this over all columns easily, but somehow I can't figure it out.
I know I could do it by 'hand', like
df[df == 'Old Value'] <- 'New value'
but this seems highly inefficient because I have 40 different values that need to be replaced.
Can someone please help me with a solution for this?

Answer 1

You can use lapply on the frame to iterate the same thing over multiple columns.
df1[] <- lapply(df1, function(z) df2$percentage[match(z, df2$scale_nr)])
df1
# sp1 sp2 sp3 sp4
# 1 NA NA 1 1
# 2 NA NA 10 30
# 3 NA NA NA NA
# 4 NA NA NA 2
The missing values are likely because of the truncated df2 in the sample data.
If you want the option to preserve the previous value if not found in df2, then you can modify that slightly:
df1[] <- lapply(df1, function(z) {
newval <- df2$percentage[match(z, df2$scale_nr)]
ifelse(is.na(newval), z, newval)
})
df1
# sp1 sp2 sp3 sp4
# 1 <NA> NA 1 1
# 2 <NA> NA 10 30
# 3 <NA> 5 <NA> <NA>
# 4 m4 NA <NA> 2
FYI, the reassignment into df1[] <- is important, in constrast with df1 <-. The difference is that lapply is going to return a list, so if you use df1 <-thendf1will no longer be adata.frame. Using df[] <-, you are telling it to replace the contents of the columns without changing the overall class of df1`.
If you need to do this on only a subset of columns, that's easy:
df1[1:3] <- lapply(df[1:3], ...)`