From 40c7d43f8cd5ccaef22f098937b985d4101368f9 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <simianer@cl.uni-heidelberg.de>
Date: Tue, 27 Sep 2011 00:10:48 +0200
Subject: some updates for hadoop streaming

---
 dtrain/hstreaming/red-avg.rb | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)
 create mode 100755 dtrain/hstreaming/red-avg.rb

(limited to 'dtrain/hstreaming/red-avg.rb')

diff --git a/dtrain/hstreaming/red-avg.rb b/dtrain/hstreaming/red-avg.rb
new file mode 100755
index 00000000..11dc0d71
--- /dev/null
+++ b/dtrain/hstreaming/red-avg.rb
@@ -0,0 +1,26 @@
+#!/usr/bin/env ruby1.9.1
+
+
+STDIN.set_encoding 'utf-8'
+
+shard_count_key = "__SHARD_COUNT__"
+
+w = {}
+c = {}
+w.default = 0
+c.default = 0
+while line = STDIN.gets
+  key, val = line.split /\t/
+  w[key] += val.to_f
+  c[key] += 1.0
+end
+
+shard_count = w["__SHARD_COUNT__"]
+
+w.each_key { |k|
+  if k == shard_count_key then next end
+  puts "#{k}\t{w[k]/shard_count}"
+}
+
+puts "#{shard_count_key}\t#{w[shard_count_key]}"
+
-- 
cgit v1.2.3