[MRG] Updated title and versions supported in contributing documentation (scikit-learn#8824)

* Changed title and updated versions supported

* Added back a line

* Fixed definition of R^2

* Fixed 80 character limit!

Author: nikitasingh981

doc/developers/contributing.rst

@@ -743,13 +743,12 @@ and the old behavior will be removed in version 0.20.
 
 .. currentmodule:: sklearn
 
-Python 3.x support
-------------------
+Python versions supported
+-------------------------
 
-All scikit-learn code should work unchanged in both Python 2.[67]
-and 3.2 or newer. Since Python 3.x is not backwards compatible,
-that may require changes to code and it certainly requires testing
-on both 2.7 and 3.2 or newer.
+All scikit-learn code should work unchanged in both Python 2.7 and 3.4 or
+newer. Since Python 3.x is not backwards compatible, that may require changes
+to code and it certainly requires testing on both 2.7 and 3.4 or newer.
 
 For most numerical algorithms, Python 3.x support is easy:
 just remember that ``print`` is a function and

sklearn/base.py

@@ -358,10 +358,10 @@ class RegressorMixin(object):
     def score(self, X, y, sample_weight=None):
         """Returns the coefficient of determination R^2 of the prediction.
 
-        The coefficient R^2 is defined as (1 - u/v), where u is the regression
-        sum of squares ((y_true - y_pred) ** 2).sum() and v is the residual
+        The coefficient R^2 is defined as (1 - u/v), where u is the residual
+        sum of squares ((y_true - y_pred) ** 2).sum() and v is the total
         sum of squares ((y_true - y_true.mean()) ** 2).sum().
-        Best possible score is 1.0 and it can be negative (because the
+        The best possible score is 1.0 and it can be negative (because the
         model can be arbitrarily worse). A constant model that always
         predicts the expected value of y, disregarding the input features,
         would get a R^2 score of 0.0.