def test_score(self):
tf = TensorCoFi(n_factors=2)
inp = [{"user": 10, "item": 100},
{"user": 10, "item": 110},
{"user": 12, "item": 120}]
inp = pd.DataFrame(inp)
tf.fit(inp)
uid = tf.data_map[tf.get_user_column()][10]
iid = tf.data_map[tf.get_item_column()][100]
tf.factors[0][uid, 0] = 0
tf.factors[0][uid, 1] = 1
tf.factors[1][iid, 0] = 1
tf.factors[1][iid, 1] = 5
self.assertEqual(0*1+1*5, tf.get_score(10, 100))
def test_refit(self):
"""
[TensorCoFi] Test fit after first fitting
"""
tf = TensorCoFi(n_factors=2)
tf.fit(self.df)
tf.fit(self.df) # Second fit
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
tf.fit(self.df) # Third fit
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
def test_ids_returns(self):
tf = TensorCoFi(n_factors=2)
inp = [{"user": 10, "item": 100},
{"user": 10, "item": 110},
{"user": 12, "item": 120}]
inp = pd.DataFrame(inp)
tf.fit(inp)
# Test the id in map
uid = tf.data_map[tf.get_user_column()][10]
iid = tf.data_map[tf.get_item_column()][100]
self.assertEquals(uid, 0)
self.assertEquals(iid, 0)
# Test number of factors
self.assertEquals(len(tf.factors[0][uid, :]), tf.number_of_factors)
self.assertEquals(len(tf.factors[1][iid, :]), tf.number_of_factors)
def test_dynamic_updates(self):
"""
TensorCoFi dynamic update
We will take a tensor cofi. Train the model, evaluate it. Then we remove all the user factors
and recompute them using the online_user_factors to check if the performance is almost the same...
"""
pyTF = PyTensorCoFi(n_factors=20, n_iterations=5, c_lambda=0.05, c_alpha=40)
evaluator = Evaluator()
tf = TensorCoFi(n_factors=2, n_iterations=100, c_lambda=0.05, c_alpha=40)
df = pd.read_csv(resource_filename(testfm.__name__, "data/movielenshead.dat"), sep="::", header=None,
names=["user", "item", "rating", "date", "title"])
training, testing = testfm.split.holdoutByRandom(df, 0.7)
users = {user: list(entries) for user, entries in training.groupby("user")["item"]}
tf.fit(training)
map1 = evaluator.evaluate_model(tf, testing) # map of the original model
#now we try to replace the original factors with on the fly computed factors
#lets iterate over the training data of items and the users
for u, items in users.items():
#user id in the tf
uid = tf.data_map[tf.get_user_column()][u] # user id
iids = [tf.data_map[tf.get_item_column()][i] for i in items] # item ids that user has seen
#original_factors = tf.factors["user"][uid]
new_factors = pyTF.online_user_factors(tf.factors[1], iids, p_param=40, lambda_param=0.05)
#replace original factors with the new factors
tf.factors[0][uid, :] = new_factors
#tf.update_user_factors(uid, new_factors)
#lets evaluate the new model with real-time updated factors
map2 = evaluator.evaluate_model(tf, testing)
#The difference should be smaller than 20%
assert abs(map1[0]-map2[0]) < 0.2*map1[0]
def test_dynamic_updates(self):
"""
TensorCoFi dynamic update
We will take a tensor cofi. Train the model, evaluate it. Then we remove all the user factors
and recompute them using the online_user_factors to check if the performance is almost the same...
"""
pyTF = PyTensorCoFi(n_factors=20,
n_iterations=5,
c_lambda=0.05,
c_alpha=40)
evaluator = Evaluator()
tf = TensorCoFi(n_factors=2,
n_iterations=100,
c_lambda=0.05,
c_alpha=40)
df = pd.read_csv(resource_filename(testfm.__name__,
"data/movielenshead.dat"),
sep="::",
header=None,
names=["user", "item", "rating", "date", "title"])
training, testing = testfm.split.holdoutByRandom(df, 0.7)
users = {
user: list(entries)
for user, entries in training.groupby("user")["item"]
}
tf.fit(training)
map1 = evaluator.evaluate_model(tf,
testing) # map of the original model
#now we try to replace the original factors with on the fly computed factors
#lets iterate over the training data of items and the users
for u, items in users.items():
#user id in the tf
uid = tf.data_map[tf.get_user_column()][u] # user id
iids = [tf.data_map[tf.get_item_column()][i]
for i in items] # item ids that user has seen
#original_factors = tf.factors["user"][uid]
new_factors = pyTF.online_user_factors(tf.factors[1],
iids,
p_param=40,
lambda_param=0.05)
#replace original factors with the new factors
tf.factors[0][uid, :] = new_factors
#tf.update_user_factors(uid, new_factors)
#lets evaluate the new model with real-time updated factors
map2 = evaluator.evaluate_model(tf, testing)
#The difference should be smaller than 20%
assert abs(map1[0] - map2[0]) < 0.2 * map1[0]
def test_refit(self):
"""
[TensorCoFi] Test fit after first fitting
"""
tf = TensorCoFi(n_factors=2)
tf.fit(self.df)
tf.fit(self.df) # Second fit
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
tf.fit(self.df) # Third fit
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
def test_ids_returns(self):
tf = TensorCoFi(n_factors=2)
inp = [{
"user": 10,
"item": 100
}, {
"user": 10,
"item": 110
}, {
"user": 12,
"item": 120
}]
inp = pd.DataFrame(inp)
tf.fit(inp)
# Test the id in map
uid = tf.data_map[tf.get_user_column()][10]
iid = tf.data_map[tf.get_item_column()][100]
self.assertEquals(uid, 0)
self.assertEquals(iid, 0)
# Test number of factors
self.assertEquals(len(tf.factors[0][uid, :]), tf.number_of_factors)
self.assertEquals(len(tf.factors[1][iid, :]), tf.number_of_factors)
def test_fit(self):
tf = TensorCoFi(n_factors=2)
tf.fit(self.df)
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
def test_score(self):
tf = TensorCoFi(n_factors=2)
inp = [{
"user": 10,
"item": 100
}, {
"user": 10,
"item": 110
}, {
"user": 12,
"item": 120
}]
inp = pd.DataFrame(inp)
tf.fit(inp)
uid = tf.data_map[tf.get_user_column()][10]
iid = tf.data_map[tf.get_item_column()][100]
tf.factors[0][uid, 0] = 0
tf.factors[0][uid, 1] = 1
tf.factors[1][iid, 0] = 1
tf.factors[1][iid, 1] = 5
self.assertEqual(0 * 1 + 1 * 5, tf.get_score(10, 100))
def test_fit(self):
tf = TensorCoFi(n_factors=2)
tf.fit(self.df)
#item and user are row vectors
self.assertEqual(len(self.df.user.unique()), tf.factors[0].shape[0])
self.assertEqual(len(self.df.item.unique()), tf.factors[1].shape[0])
from testfm.evaluation.evaluator import Evaluator
from pkg_resources import resource_filename
from testfm.evaluation.parameter_tuning import ParameterTuning
if __name__ == "__main__":
eval = Evaluator(
) # Call this before loading the data to save memory (fork of process takes place)
# Prepare the data
df = pd.read_csv(resource_filename(testfm.__name__,
'data/movielenshead.dat'),
sep="::",
header=None,
names=['user', 'item', 'rating', 'date', 'title'])
print df.head()
training, testing = testfm.split.holdoutByRandom(df, 0.9)
print "Tuning the parameters."
tr, validation = testfm.split.holdoutByRandom(training, 0.7)
pt = ParameterTuning()
pt.set_max_iterations(100)
pt.set_z_value(90)
tf_params = pt.get_best_params(TensorCoFi, tr, validation)
print tf_params
tf = TensorCoFi()
tf.set_params(**tf_params)
tf.fit(training)
print tf.get_name().ljust(50),
print eval.evaluate_model(tf, testing, all_items=training.item.unique())
