I have a dataframe with lots of columns, including four that represents sub-categories.
data.frame(site_name=c("bla","blo","blu"), page_1=c(NA,NA,NA), page_2=c(NA,"detail_1","detail_2"), page_3=c("hello", "hola", NA), page_4=c(NA,NA,NA))
site_name page_1 page_2 page_3 page_4
1 bla NA <NA> hello NA
2 blo NA detail_1 hola NA
3 blu NA detail_2 <NA> NA
I would like to replace all these 'page_x' columns with one single column, showing only the right-most non-NA level of detail. There are lots of row-wise questions, but I cannot find one working with NSE and Reduce.
In the example above that would be:
site_name page_1 page_2 page_3 page_4 page
1 bla NA <NA> hello NA hello
2 blo NA detail_1 hola NA hola
3 blu NA detail_2 <NA> NA detail_2
I thought it would be a good example to use Reduce over each row. I have a version that kind of work on one single row, but I cannot find a way to apply it to mutate_.
Reduce(function(prec,col){ifelse(!is.na(row[col]), row[col], prec)},
grep("^page",names(row),value=T), NA)}
This does not work (as I guess the 'row' object is not really passed this way):
mutate_(.dots = ~Reduce(function(prec,col){ifelse(!is.na(row[col]), row[col], prec)},
grep("^page",names(row),value=T), NA)}
