How can I optimize this array.each in ruby?

Question

I'm trying to optimize this bit of code and I thought I'd throw it out here to see if anyone has any ideas. I've been testing the performance using Benchmark.measure {...} and wrapping the entire function.

Yes, the item has a date that is a Ruby Date

 puts Benchmark.measure {
 grouped_items
 }
 def grouped_items
 unless @grouped
 @grouped = {}
 @items.each do |item|
 key = item.date
 if @grouped.has_key?(key)
 @grouped[key] << item
 else
 @grouped[key] = [item]
 end
 end
 end
 @grouped
 end

Thanks for any insight you care to share.

Edit #1: My first optimization. I was aiming for a slightly more succinct function and gained a .2 seconds reduction in time to process 100_000 items.

 def grouped_items
 unless @grouped
 @grouped = Hash.new { |h, k| h[k] = [] }
 @items.map {|item| @grouped[item.date] << item }
 end
 @grouped
 end

Edit #2: My second iteration with more or less the same performance profile but with much less code.

def grouped_items
 @grouped ||= @items.group_by { |item| item.date }
end

Edit #3: An alternative way to do the same thing above.

def grouped_items
 @grouped ||= @items.group_by &:date
end

• rmontgomery429: Nope. I'll check that out.
• Kev: In future can you flag for a mod to migrate, please don't cross-post. Thanks.
• Andrew Grimm: Did you try profiling the code?

Answer 1

Are you aware of Enumerable#group_by?

def grouped_items
 @items.group_by{|item| item.date}
end

• rmontgomery429: It more or less performs the same as the first edit (see above) but with much less code.
• rmontgomery429: I just came back across this and wanted to comment again how awesome this solution is. :)
• Aaron Cronin: Just to add, you could even use symbol-to-proc here since you are grouping by an immediate method call: @items.group_by &:date.
• Mark Thomas: Indeed, as the OP discovered and showed in his Edit #3.

Answer 2

What about this?

@items.group_by { |item| item.effective_date }

Benchmarks for the curious:

Ryan v1 : 1.080000 0.000000 1.080000 ( 1.077599)
Ryan v2 : 0.620000 0.000000 0.620000 ( 0.622756)
group_by: 0.580000 0.000000 0.580000 ( 0.576531)