class CollabFiltTest(unittest.TestCase):
def setUp(self):
log.debug("setup...")
self.filtNA = Filt(missingaszero=False)
self.filtNA.addUser('u1', {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.filtNA.addUser('u2', {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.filtNA.addUser('u3', {'i1':2.5})
self.filtNA.addUser('u4', {'i1':5.0, 'i3':3.0})
self.filtNA.addUser('u5', {'i1':4.0, 'i2':3.0, 'i3':2.0})
self.filtZE = Filt(missingaszero=True)
self.filtZE.addUser('u1', {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.filtZE.addUser('u2', {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.filtZE.addUser('u3', {'i1':2.5})
self.filtZE.addUser('u4', {'i1':5.0, 'i3':3.0})
self.filtZE.addUser('u5', {'i1':4.0, 'i2':3.0, 'i3':2.0})
def tearDown(self):
log.debug("teardown...")
def test_GetUserCount(self):
self.assertEqual(self.filtNA.getUserCount(), 5)
self.assertEqual(self.filtZE.getUserCount(), 5)
def test_GetItemCount(self):
self.assertEqual(self.filtNA.getItemCount(), 3)
self.assertEqual(self.filtZE.getItemCount(), 3)
def test_GetUserRatings(self):
ratings = self.filtNA.getUserRatings('u4')
self.assertEqual(ratings, {'i1':5.0, 'i3':3.0})
ratings = self.filtZE.getUserRatings('u4')
self.assertEqual(ratings, {'i1':5.0, 'i2':0, 'i3':3.0})
def test_GetUserRatingCounts(self):
counts = self.filtNA.getUserRatingCounts()
self.assertEqual(set(counts), set([('u1', 3), ('u4', 2), ('u5', 3), ('u3', 1), ('u2', 3)]))
counts = self.filtZE.getUserRatingCounts()
self.assertEqual(set(counts), set([('u1', 3), ('u2', 3), ('u3', 3), ('u4', 3), ('u5', 3)]))
def test_GetItemRatings(self):
ratings = self.filtNA.getItemRatings('i2')
self.assertEqual(ratings, {'u1': 3.0, 'u2': 2.5, 'u5': 3.0})
ratings = self.filtZE.getItemRatings('i2')
self.assertEqual(ratings, {'u1': 3.0, 'u2': 2.5, 'u3':0, 'u4':0, 'u5': 3.0})
def test_GetItemRatingCounts(self):
counts = self.filtNA.getItemRatingCounts()
self.assertEqual(set(counts), set([('i1', 5), ('i3', 4), ('i2', 3)]))
counts = self.filtZE.getItemRatingCounts()
self.assertEqual(set(counts), set([('i1', 5), ('i3', 5), ('i2', 5)]))
def test_GetRatingMean(self):
mean = self.filtNA.getRatingMean({'u1': 3.0, 'u2': 2.5, 'u3':0, 'u4':0, 'u5': 3.0})
self.assertEqual(mean, 1.7)
def test_Euclid (self):
# NA
sim = self.filtNA.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.assertAlmostEqual(sim, 0.203, delta=0.001)
sim = self.filtNA.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.5})
self.assertAlmostEqual(sim, 0.286, delta=0.001)
sim = self.filtNA.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i3':3.0})
self.assertAlmostEqual(sim, 0.667, delta=0.001)
sim = self.filtNA.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':4.0, 'i2':3.0, 'i3':2.0})
self.assertAlmostEqual(sim, 0.472, delta=0.001)
# ZE
sim = self.filtZE.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtZE.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.assertAlmostEqual(sim, 0.203, delta=0.001)
sim = self.filtZE.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.5})
self.assertAlmostEqual(sim, 0.177, delta=0.001)
sim = self.filtZE.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i3':3.0})
self.assertAlmostEqual(sim, 0.247, delta=0.001)
sim = self.filtZE.euclid({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':4.0, 'i2':3.0, 'i3':2.0})
self.assertAlmostEqual(sim, 0.472, delta=0.001)
def test_Pearson (self):
sim = self.filtNA.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.assertAlmostEqual(sim, -0.764, delta=0.001)
sim = self.filtNA.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.5})
self.assertIsNone(sim)
sim = self.filtNA.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i3':3.0})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':4.0, 'i2':3.0, 'i3':2.0})
self.assertAlmostEqual(sim, 0.945, delta=0.001)
# ZE
sim = self.filtZE.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i2':3.0, 'i3':2.5})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtZE.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.0, 'i2':2.5, 'i3':5.0})
self.assertAlmostEqual(sim, -0.764, delta=0.001)
sim = self.filtZE.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':2.5})
self.assertAlmostEqual(sim, 0.982, delta=0.001)
sim = self.filtZE.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':5.0, 'i3':3.0})
self.assertAlmostEqual(sim, 0.676, delta=0.001)
sim = self.filtZE.pearson({'i1':5.0, 'i2':3.0, 'i3':2.5}, {'i1':4.0, 'i2':3.0, 'i3':2.0})
self.assertAlmostEqual(sim, 0.945, delta=0.001)
def test_SimilarUsers(self):
# NA
sim = self.filtNA.similarUsers({'i1':5.0, 'i2':3.0, 'i3':2.5}, n=5, sim='euclid')
self.assertEqual(sim[0][0], 'u1')
self.assertAlmostEqual(sim[0][1], 1.0, delta=0.001)
self.assertEqual(sim[1][0], 'u4')
self.assertAlmostEqual(sim[1][1], 0.667, delta=0.001)
self.assertEqual(sim[2][0], 'u5')
self.assertAlmostEqual(sim[2][1], 0.472, delta=0.001)
self.assertEqual(sim[3][0], 'u3')
self.assertAlmostEqual(sim[3][1], 0.286, delta=0.001)
self.assertEqual(sim[4][0], 'u2')
self.assertAlmostEqual(sim[4][1], 0.203, delta=0.001)
# ZE
sim = self.filtZE.similarUsers({'i1':5.0, 'i2':3.0, 'i3':2.5}, n=5, sim='euclid')
self.assertEqual(sim[0][0], 'u1')
self.assertAlmostEqual(sim[0][1], 1.0, delta=0.001)
self.assertEqual(sim[1][0], 'u5')
self.assertAlmostEqual(sim[1][1], 0.472, delta=0.001)
self.assertEqual(sim[2][0], 'u4')
self.assertAlmostEqual(sim[2][1], 0.247, delta=0.001)
self.assertEqual(sim[3][0], 'u2')
self.assertAlmostEqual(sim[3][1], 0.203, delta=0.001)
self.assertEqual(sim[4][0], 'u3')
self.assertAlmostEqual(sim[4][1], 0.177, delta=0.001)
def test_predictRatings (self):
# NA
rat = self.filtNA.predictRatings({'i1':4.0})
self.assertEquals(len(rat), 2)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 2.909, delta=0.001)
self.assertEqual(rat[1][0], 'i3')
self.assertAlmostEqual(rat[1][1], 2.750, delta=0.001)
# NA m=1
rat = self.filtNA.predictRatings({'i1':4.0}, m=1)
self.assertEquals(len(rat), 1)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 2.909, delta=0.001)
# NA n=3
rat = self.filtNA.predictRatings({'i1':4.0}, n=3)
self.assertEquals(len(rat), 2)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 3.0, delta=0.001)
self.assertEqual(rat[1][0], 'i3')
self.assertAlmostEqual(rat[1][1], 2.375, delta=0.001)
# DictReader expects first row to define fieldnames
reader = csv.DictReader(f, delimiter=';', quotechar='"')
for line in reader:
books[line['ISBN']] = line['Book-Title'], line['Book-Author']
# Load ratings from input file
filt = Filt(missingaszero=False)
with open(input, 'rb') as f:
# DictReader expects first row to define fieldnames
reader = csv.DictReader(f, delimiter=';', quotechar='"')
for line in reader:
filt.addUserRatings(line['User-ID'], {line['ISBN']: line['Book-Rating']})
print ("User count: %d" %filt.getUserCount())
print ("Item count: %d" % filt.getItemCount())
# "042505313X";"Dune";"Frank Herbert"
# "083760463X";"The Martian Way and Other Stories";"Isaac Asimov"
# "0425042367";"Podkayne of Mars";"Robert A. Heinlein"
# "0425043770";"Stranger Strg Lnd";"Robert A. Heinlein"
# "0425043797";"Dune Messiah";"Frank Herbert"
# "0446354872";"Batman: The Novelization"
user1 = {'042505313X':10, '083760463X':7, "0425042367":8, "0425043770":6, "0425043797":10, '0446354872': 7.0}
print ("\nUser\n====")
mapISBN ([(isbn, rating) for (isbn, rating) in user1.items()])
topnitems = 10
print ("\nEuclid\n======")
print ("Top estimated ratings for user:")
predicted = filt.predictRatings(user1, m=topnitems, sim='euclid')
# DictReader expects first row to define fieldnames
reader = csv.DictReader(f, delimiter=';', quotechar='"')
for line in reader:
books[line['ISBN']] = line['Book-Title'], line['Book-Author']
# Load ratings from input file
filt = Filt(missingaszero=False)
with open(input, 'rb') as f:
# DictReader expects first row to define fieldnames
reader = csv.DictReader(f, delimiter=';', quotechar='"')
for line in reader:
filt.addUserRatings(line['User-ID'],
{line['ISBN']: line['Book-Rating']})
print("User count: %d" % filt.getUserCount())
print("Item count: %d" % filt.getItemCount())
# "042505313X";"Dune";"Frank Herbert"
# "083760463X";"The Martian Way and Other Stories";"Isaac Asimov"
# "0425042367";"Podkayne of Mars";"Robert A. Heinlein"
# "0425043770";"Stranger Strg Lnd";"Robert A. Heinlein"
# "0425043797";"Dune Messiah";"Frank Herbert"
# "0446354872";"Batman: The Novelization"
user1 = {
'042505313X': 10,
'083760463X': 7,
"0425042367": 8,
"0425043770": 6,
"0425043797": 10,
'0446354872': 7.0
}
class CollabFiltTest(unittest.TestCase):
def setUp(self):
log.debug("setup...")
self.filtNA = Filt(missingaszero=False)
self.filtNA.addUser('u1', {'i1': 5.0, 'i2': 3.0, 'i3': 2.5})
self.filtNA.addUser('u2', {'i1': 2.0, 'i2': 2.5, 'i3': 5.0})
self.filtNA.addUser('u3', {'i1': 2.5})
self.filtNA.addUser('u4', {'i1': 5.0, 'i3': 3.0})
self.filtNA.addUser('u5', {'i1': 4.0, 'i2': 3.0, 'i3': 2.0})
self.filtZE = Filt(missingaszero=True)
self.filtZE.addUser('u1', {'i1': 5.0, 'i2': 3.0, 'i3': 2.5})
self.filtZE.addUser('u2', {'i1': 2.0, 'i2': 2.5, 'i3': 5.0})
self.filtZE.addUser('u3', {'i1': 2.5})
self.filtZE.addUser('u4', {'i1': 5.0, 'i3': 3.0})
self.filtZE.addUser('u5', {'i1': 4.0, 'i2': 3.0, 'i3': 2.0})
def tearDown(self):
log.debug("teardown...")
def test_GetUserCount(self):
self.assertEqual(self.filtNA.getUserCount(), 5)
self.assertEqual(self.filtZE.getUserCount(), 5)
def test_GetItemCount(self):
self.assertEqual(self.filtNA.getItemCount(), 3)
self.assertEqual(self.filtZE.getItemCount(), 3)
def test_GetUserRatings(self):
ratings = self.filtNA.getUserRatings('u4')
self.assertEqual(ratings, {'i1': 5.0, 'i3': 3.0})
ratings = self.filtZE.getUserRatings('u4')
self.assertEqual(ratings, {'i1': 5.0, 'i2': 0, 'i3': 3.0})
def test_GetUserRatingCounts(self):
counts = self.filtNA.getUserRatingCounts()
self.assertEqual(
set(counts),
set([('u1', 3), ('u4', 2), ('u5', 3), ('u3', 1), ('u2', 3)]))
counts = self.filtZE.getUserRatingCounts()
self.assertEqual(
set(counts),
set([('u1', 3), ('u2', 3), ('u3', 3), ('u4', 3), ('u5', 3)]))
def test_GetItemRatings(self):
ratings = self.filtNA.getItemRatings('i2')
self.assertEqual(ratings, {'u1': 3.0, 'u2': 2.5, 'u5': 3.0})
ratings = self.filtZE.getItemRatings('i2')
self.assertEqual(ratings, {
'u1': 3.0,
'u2': 2.5,
'u3': 0,
'u4': 0,
'u5': 3.0
})
def test_GetItemRatingCounts(self):
counts = self.filtNA.getItemRatingCounts()
self.assertEqual(set(counts), set([('i1', 5), ('i3', 4), ('i2', 3)]))
counts = self.filtZE.getItemRatingCounts()
self.assertEqual(set(counts), set([('i1', 5), ('i3', 5), ('i2', 5)]))
def test_GetRatingMean(self):
mean = self.filtNA.getRatingMean({
'u1': 3.0,
'u2': 2.5,
'u3': 0,
'u4': 0,
'u5': 3.0
})
self.assertEqual(mean, 1.7)
def test_Euclid(self):
# NA
sim = self.filtNA.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 2.0,
'i2': 2.5,
'i3': 5.0
})
self.assertAlmostEqual(sim, 0.203, delta=0.001)
sim = self.filtNA.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {'i1': 2.5})
self.assertAlmostEqual(sim, 0.286, delta=0.001)
sim = self.filtNA.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i3': 3.0
})
self.assertAlmostEqual(sim, 0.667, delta=0.001)
sim = self.filtNA.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 4.0,
'i2': 3.0,
'i3': 2.0
})
self.assertAlmostEqual(sim, 0.472, delta=0.001)
# ZE
sim = self.filtZE.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtZE.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 2.0,
'i2': 2.5,
'i3': 5.0
})
self.assertAlmostEqual(sim, 0.203, delta=0.001)
sim = self.filtZE.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {'i1': 2.5})
self.assertAlmostEqual(sim, 0.177, delta=0.001)
sim = self.filtZE.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i3': 3.0
})
self.assertAlmostEqual(sim, 0.247, delta=0.001)
sim = self.filtZE.euclid({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 4.0,
'i2': 3.0,
'i3': 2.0
})
self.assertAlmostEqual(sim, 0.472, delta=0.001)
def test_Pearson(self):
sim = self.filtNA.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 2.0,
'i2': 2.5,
'i3': 5.0
})
self.assertAlmostEqual(sim, -0.764, delta=0.001)
sim = self.filtNA.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {'i1': 2.5})
self.assertIsNone(sim)
sim = self.filtNA.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i3': 3.0
})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtNA.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 4.0,
'i2': 3.0,
'i3': 2.0
})
self.assertAlmostEqual(sim, 0.945, delta=0.001)
# ZE
sim = self.filtZE.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
})
self.assertAlmostEqual(sim, 1.0, delta=0.001)
sim = self.filtZE.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 2.0,
'i2': 2.5,
'i3': 5.0
})
self.assertAlmostEqual(sim, -0.764, delta=0.001)
sim = self.filtZE.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {'i1': 2.5})
self.assertAlmostEqual(sim, 0.982, delta=0.001)
sim = self.filtZE.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 5.0,
'i3': 3.0
})
self.assertAlmostEqual(sim, 0.676, delta=0.001)
sim = self.filtZE.pearson({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
}, {
'i1': 4.0,
'i2': 3.0,
'i3': 2.0
})
self.assertAlmostEqual(sim, 0.945, delta=0.001)
def test_SimilarUsers(self):
# NA
sim = self.filtNA.similarUsers({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
},
n=5,
sim='euclid')
self.assertEqual(sim[0][0], 'u1')
self.assertAlmostEqual(sim[0][1], 1.0, delta=0.001)
self.assertEqual(sim[1][0], 'u4')
self.assertAlmostEqual(sim[1][1], 0.667, delta=0.001)
self.assertEqual(sim[2][0], 'u5')
self.assertAlmostEqual(sim[2][1], 0.472, delta=0.001)
self.assertEqual(sim[3][0], 'u3')
self.assertAlmostEqual(sim[3][1], 0.286, delta=0.001)
self.assertEqual(sim[4][0], 'u2')
self.assertAlmostEqual(sim[4][1], 0.203, delta=0.001)
# ZE
sim = self.filtZE.similarUsers({
'i1': 5.0,
'i2': 3.0,
'i3': 2.5
},
n=5,
sim='euclid')
self.assertEqual(sim[0][0], 'u1')
self.assertAlmostEqual(sim[0][1], 1.0, delta=0.001)
self.assertEqual(sim[1][0], 'u5')
self.assertAlmostEqual(sim[1][1], 0.472, delta=0.001)
self.assertEqual(sim[2][0], 'u4')
self.assertAlmostEqual(sim[2][1], 0.247, delta=0.001)
self.assertEqual(sim[3][0], 'u2')
self.assertAlmostEqual(sim[3][1], 0.203, delta=0.001)
self.assertEqual(sim[4][0], 'u3')
self.assertAlmostEqual(sim[4][1], 0.177, delta=0.001)
def test_predictRatings(self):
# NA
rat = self.filtNA.predictRatings({'i1': 4.0})
self.assertEquals(len(rat), 2)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 2.909, delta=0.001)
self.assertEqual(rat[1][0], 'i3')
self.assertAlmostEqual(rat[1][1], 2.750, delta=0.001)
# NA m=1
rat = self.filtNA.predictRatings({'i1': 4.0}, m=1)
self.assertEquals(len(rat), 1)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 2.909, delta=0.001)
# NA n=3
rat = self.filtNA.predictRatings({'i1': 4.0}, n=3)
self.assertEquals(len(rat), 2)
self.assertEqual(rat[0][0], 'i2')
self.assertAlmostEqual(rat[0][1], 3.0, delta=0.001)
self.assertEqual(rat[1][0], 'i3')
self.assertAlmostEqual(rat[1][1], 2.375, delta=0.001)