From 99468e022ed1b0854b7dcc70eb00bba6940e5bb8 Mon Sep 17 00:00:00 2001
From: "trevor.cohn" <trevor.cohn@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Mon, 12 Jul 2010 16:38:52 +0000
Subject: Code for evaluating clusterings

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@222 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/evaluation/evaluate_entropy.py | 46 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)
 create mode 100644 gi/evaluation/evaluate_entropy.py

(limited to 'gi/evaluation/evaluate_entropy.py')

diff --git a/gi/evaluation/evaluate_entropy.py b/gi/evaluation/evaluate_entropy.py
new file mode 100644
index 00000000..88533544
--- /dev/null
+++ b/gi/evaluation/evaluate_entropy.py
@@ -0,0 +1,46 @@
+#!/usr/bin/env python
+
+import sys, math, itertools
+
+ginfile = open(sys.argv[1])
+pinfile = open(sys.argv[2])
+
+# evaluating: H(G | P) = sum_{g,p} p(g,p) log { p(p) / p(g,p) }
+#                      = sum_{g,p} c(g,p)/N { log c(p) - log N - log c(g,p) + log N }
+#                      = 1/N sum_{g,p} c(g,p) { log c(p) - log c(g,p) }
+# where G = gold, P = predicted, N = number of events
+
+N = 0
+gold_frequencies = {}
+predict_frequencies = {}
+joint_frequencies = {}
+
+for gline, pline in itertools.izip(ginfile, pinfile):
+    gparts = gline.split('||| ')[1].split()
+    pparts = pline.split('||| ')[1].split()
+    assert len(gparts) == len(pparts)
+
+    for gpart, ppart in zip(gparts, pparts):
+        gtag = gpart.split(':',1)[1]
+        ptag = ppart.split(':',1)[1]
+
+        joint_frequencies.setdefault((gtag, ptag), 0)
+        joint_frequencies[gtag,ptag] += 1
+
+        predict_frequencies.setdefault(ptag, 0)
+        predict_frequencies[ptag] += 1
+
+        gold_frequencies.setdefault(gtag, 0)
+        gold_frequencies[gtag] += 1
+
+        N += 1
+
+hg2p = 0
+hp2g = 0
+for (gtag, ptag), cgp in joint_frequencies.items():
+    hp2g += cgp * (math.log(predict_frequencies[ptag], 2) - math.log(cgp, 2))
+    hg2p += cgp * (math.log(gold_frequencies[gtag], 2) - math.log(cgp, 2))
+hg2p /= N
+hp2g /= N
+
+print 'H(P|G)', hg2p, 'H(G|P)', hp2g, 'VI', hg2p + hp2g
-- 
cgit v1.2.3