Added average caches. Completed Unit Tests. Fixes #14 and fixes #3

Netflix.py

@@ -9,7 +9,14 @@
 CACHE_PROBE = 0
 CACHE_GENERAL = 1
 
-cache_probe = {}
+cache_probe = None
+cache_movie_averages = None
+cache_cust_averages = None
+cache_num_ratings = None
+cache_movie_decade_avg = None
+
+PROBE_MEAN = 3.7
+AVG_NUM_RATINGS = 5654.5
 
 class Prediction:
   def __init__(self, movie, custids, predictions):
@@ -23,7 +30,7 @@ def display(self, writer):
       writer.write(str(round(rating, 1)) + "\n") 
 
   def calculate_rmse(self) :
-    return rmse(self.predictions, real_ratings())
+    return rmse(self.predictions, self.real_ratings())
 
   def real_ratings(self) :
     return map(lambda x: get_cache_probe()[self.movie][x], self.custids)
@@ -35,7 +42,7 @@ def netflix_solve(r, w) :
   r is a reader
   w is a writer
   """
-  get_cache_probe() # Load probe cache if it has not been loaded yet
+  netflix_load_caches()
 
   assert(len(cache_probe) > 0)
 
@@ -64,11 +71,29 @@ def netflix_solve(r, w) :
 def netflix_predict(movie, custids) :
   assert(movie > 0)
   assert(len(custids) > 0)
-  prediction = Prediction(movie, custids, len(custids) * [3.7])
+  netflix_load_caches()
+  predicted_ratings = []
+  for custid in custids:
+    rating = cache_movie_decade_avg[movie] + (cache_movie_averages[movie] - cache_movie_decade_avg[movie]) + (cache_cust_averages[custid] - PROBE_MEAN)
+    # rating = rating / 4
+    rating = max(min(rating, 5.0), 1.0) # Ratings can only be between 1.0 and 5.0
+    predicted_ratings.append(rating)
+  prediction = Prediction(movie, custids, predicted_ratings)
   assert(prediction is not None)
   return prediction
 
   # with open("irvin-user_avg_json", "r")
+  # 
+def netflix_load_caches():
+  get_cache_probe()
+  global cache_movie_averages
+  global cache_cust_averages
+  global cache_num_ratings
+  global cache_movie_decade_avg
+  cache_movie_averages = cache_movie_averages or netflix_load_cache(CACHE_GENERAL, "ericweb2-movieAveragesOneLine.txt")
+  cache_cust_averages = cache_cust_averages or netflix_load_cache(CACHE_GENERAL, "ericweb2-custAveragesOneLine.txt")
+  cache_num_ratings = cache_num_ratings or netflix_load_cache(CACHE_GENERAL, "ericweb2-numRatingsOneLine.txt")
+  cache_movie_decade_avg = cache_movie_decade_avg or netflix_load_cache(CACHE_GENERAL, "ericweb2-movieDecadeAvgRatingOneLine.txt")
 
 def netflix_load_cache(cache_type, path):
   """

TestNetflix.py

@@ -10,7 +10,7 @@
 import io
 import unittest
 
-from Netflix import netflix_solve, netflix_predict, rmse, netflix_load_cache, set_cache_probe, get_cache_probe, Prediction, CACHE_PROBE, CACHE_GENERAL
+from Netflix import netflix_solve, netflix_predict, rmse, netflix_load_cache, netflix_load_caches, set_cache_probe, get_cache_probe, Prediction, CACHE_PROBE, CACHE_GENERAL
 
 class TestNetflix (unittest.TestCase) :
   # ---
@@ -34,23 +34,29 @@ def test_solve3 (self) :
     netflix_solve(r, w)
     self.assertEqual(w.getvalue(), "10007:\n2.0\n2.7\n2.2\nRMSE: 1.6254316644013882\n")
 
+  def test_solve4 (self) :
+    r = io.StringIO("3676:\n1982226\n1495499\n1669842\n2639376\n1982597\n1426939\n")
+    w = io.StringIO()
+    netflix_solve(r, w)
+    self.assertEqual(w.getvalue(), "3676:\n3.7\n3.6\n4.5\n2.9\n4.2\n4.2\nRMSE: 0.6597351087461958\n")
+
   # ---
   # netflix_predict
   # ---
   def test_predict(self) :
-    prediction = netflix_predict(1, (4,8,10,12))
-    self.assertEqual(prediction.movie, 1)
-    self.assertEqual(prediction.predictions, 4 * [3.7])
+    prediction = netflix_predict(3674, (1100257, 347434, 2472269))
+    self.assertEqual(prediction.movie, 3674)
+    self.assertEqual(prediction.predictions, [4.21234183018522, 3.377376155128011, 3.660141657622665])
 
   def test_predict2(self) :
-    prediction = netflix_predict(2, (1,2,3,4))
+    prediction = netflix_predict(3675, (1982589, 1376081, 2309102))
     s = io.StringIO()
-    self.assertEqual(prediction.movie, 2)
+    self.assertEqual(prediction.movie, 3675)
     prediction.display(s)
-    self.assertEqual(s.getvalue(), "2:\n3.7\n3.7\n3.7\n3.7\n")
+    self.assertEqual(s.getvalue(), "3675:\n2.3\n3.1\n2.9\n")
 
   def test_predict3(self) :
-    prediction = netflix_predict(1, (1,2,3))
+    prediction = netflix_predict(3674, (1100257, 347434, 2472269))
     self.assertTrue(type(prediction), Prediction)
 
   def test_prediction_display(self) :
@@ -65,6 +71,10 @@ def test_prediction_display2(self) :
     prediction.display(s)
     self.assertEqual(s.getvalue(), "2:\n2\n3\n")
 
+  def test_prediction_real_ratings(self) :
+    prediction = Prediction(3675, (1982589, 1376081, 2309102), [4,5,2])
+    self.assertEqual(list(prediction.real_ratings()), [4, 3, 1])
+
   def test_prediction_rmse(self) :
     set_cache_probe({1: {1: 3, 4:2, 5:5}})
     prediction = Prediction(1, [1,4,5], [3.7, 2.1, 4.2])
@@ -106,6 +116,10 @@ def test_get_cache_probe(self) :
     self.assertEqual(len(cache[10]), 2)
     self.assertEqual(cache[10][1531863], 3)
 
+  def test_load_caches(self) :
+    netflix_load_caches()
+    self.assertTrue(get_cache_probe is not None)
+
   # ---
   # rmse
   # ---