$2 ? ... will give the wrong result [...] This can be fixed using $2 ne ''

Well, it seems to me from the original post that the file will always have both columns; if the match doesn't succeed then the line is not valid and (maybe?) shouldn't be considered.

The code becomes (note that map in void context is almost always considerably slower than for):

# assuming @i = map "name$_" 1..5
/(\S+)\s+(-?\d+(?:.\d+)?)/and$a{$1}=$2 for<DATA>;
push@o,exists$a{$_}?$a{$_}:999 for@i;
[download]

Remember, this simply ignores lines in the file that don't have both columns. It also ignores a second decimal (and everything after) in the second column. It all depends on how lenient we want to be of bad data.

bbfu
Seasons don't fear The Reaper.
Nor do the wind, the sun, and the rain.
We can be like they are.