From 0d55a9fae74edf990a087463a52b81ef196862a2 Mon Sep 17 00:00:00 2001
From: Naftali Harris <naftaliharris@gmail.com>
Date: Tue, 29 Jul 2014 23:46:30 -0700
Subject: [PATCH] Avoid numerical instability

This avoids basically doing 1 - 1, for example:

>>> from math import exp
>>> margin = -40
>>> 1 - 1 / (1 + exp(margin))
0.0
>>> exp(margin) / (1 + exp(margin))
4.248354255291589e-18
>>>
---
 python/pyspark/mllib/classification.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/python/pyspark/mllib/classification.py b/python/pyspark/mllib/classification.py
index 9e28dfbb9145d..2bbb9c3fca315 100644
--- a/python/pyspark/mllib/classification.py
+++ b/python/pyspark/mllib/classification.py
@@ -66,7 +66,8 @@ def predict(self, x):
         if margin > 0:
             prob = 1 / (1 + exp(-margin))
         else:
-            prob = 1 - 1 / (1 + exp(margin))
+            exp_margin = exp(margin)
+            prob = exp_margin / (1 + exp_margin)
         return 1 if prob > 0.5 else 0