R: Replacing values conditional on another column AND matching variable names - r

Question

My overall goal is to assign values to a new variable from one of several variables with specific string matches conditional on the value of another variable. More specifically:
I am trying to add many columns to a data frame where each of the given new columns (e.g. 'foo') takes on the value of one of two columns already in the data frame and whose names begin with the same string and end with one of two suffixes (e.g. 'foo.2009' and 'foo.2014') conditional on the value of another column (e.g. 'year'). The data frame also contains columns unrelated to this operation and these are identified by their lack of suffixes (e.g. 'other_example' do not end in '.2009' or '.2014') and I have created a vector of the names of the new columns. In the below example data, I want to assign values to foo from foo.2014 if year >=2014 and from foo.2009 if year < 2014.
# Original data frame
df <- data.frame( foo.2009 = seq(1,3),
foo.2014 = seq(5,7),
foo = NA,
bar = NA,
other_example = seq(20,22),
year = c(2014,2009,2014))
print(df)
# The vector of variable names ending in '.####`
names <- c("foo")
# Target data frame
df$foo <- c(5,2,7)
print(df)
In my real data, I have many variables (e.g. bar) similar to foo where I want bar == bar.2014 if year >= 2014 and bar == bar.2009 if year < 2014. I am therefore trying to develop a solution where I can loop through (or use vectorized operations on) a vector of variable names (e.g. names) for an arbitrarily large number of variables where I want to replace the values:
# The vector of variable names ending in `.####`
names <- c("foo","bar")
# Original data frame
df <- data.frame( foo.2009 = seq(1,3),
foo.2014 = seq(5,7),
bar.2009 = seq(8,10),
bar.2014 = rep(5,3),
foo = NA,
bar = NA,
other_example = seq(20,22),
year = c(2014,2009,2014))
df
# Target data frame
df$foo <- c(5,2,7)
df$bar <- c(5,9,5)
df
I am particularly having trouble with the need to evaluate multiple strings comprising variable names in a loop or using a vectorized approach. An attempt is below using dplyr::mutate() to add the variables then assign them values. Below is the same data as above but an example of what an additional variable to recode would look like.
library(dplyr)
for (i in names){
var09 <- paste0(i, ".2009")
var14 <- paste0(i, ".2014")
dplyr::mutate_(df,
i = ifelse(df$year < 2010,
paste0("df$",i, ".2009"),
paste0("df$",i, ".2014")))}

Answer 1

We can loop through the sequence in base R
nm1 <- c("foo\\.\\d+", "bar\\.\\d+")
nm2 <- c("foo", "bar")
for(j in seq_along(nm1)){
sub1 <- df[grep(nm1[j], names(df))]
df[[nm2[j]]] <- ifelse(df$year < 2010, sub1[[1]], sub1[[2]])
}
df
# foo.2009 foo.2014 bar.2009 bar.2014 foo bar other_example year
#1 1 5 8 5 5 5 20 2014
#2 2 6 9 5 2 9 21 2009
#3 3 7 10 5 7 5 22 2014