Wondering if filter would be more performant.

Exactly what I did.

But since and empty array may be passed as an argument, this becomes mandatory, otherwise you'd get and error.

Yes, thank you very much!

Normally I'd say yes, giving parseInt a radix of 10 is best form. However, you really only have to worry about specifying a radix if the string could possibly have a leading "0x" or "0". E.g.

parseInt("0x10"); // -> 16
parseInt("010"); // -> 10 or 8, depending on implementation

Both of those are impossible in this case: we know that the string is a number in string form and the digits are sorted in descending order, so it can't possibly begin with "0x" or "0" (unless the number is 0, and then the radix is irrelevant).

I prefer +n anyway in most cases, because you can test that the result isn't NaN. parseInt and parseFloat provide no error handling whatsoever.

toFixed does round.

Can anyone explain? Thanks :-)

I submitted this solution and received an error.
Anyone ideas?