ENH Improve code readability

This is likely to be slightly slower, but it's easier to explain and this code is to teach the concepts, not raw speed.

Author: luispedro

ch08/corrneighbours.py

@@ -18,21 +18,25 @@ def predict(otrain):
 norm=NormalizePositive()
 train=norm.fit_transform(otrain.T).T
 
-proximity=distance.pdist(binary, 'correlation')
-proximity=distance.squareform(proximity)
+dists=distance.pdist(binary, 'correlation')
+dists=distance.squareform(dists)
 
-neighbors=proximity.argsort(axis=1)
+neighbors=dists.argsort(axis=1)
 filled=train.copy()
 foruinrange(filled.shape[0]):
+# n_u are the neighbors of user
 n_u=neighbors[u, 1:]
-t_u=train[n_u].T
-b_u=binary[n_u].T
 forminrange(filled.shape[1]):
-revs=t_u[m]
-brevs=b_u[m]
-revs=revs[brevs]
+# This code could be faster using numpy indexing trickery as the
+# cost of readibility (this is left as an exercise to the reader):
+revs= [train[neigh, m]
+forneighinn_u
+ifbinary[neigh, m]]
 iflen(revs):
-revs=revs[:len(revs)//2+1]
+n=len(revs)
+n//=2
+n+=1
+revs=revs[:n]
 filled[u,m] =revs.mean()
 
 returnnorm.inverse_transform(filled.T).T