def test_bpmf_tensor(self):
np.random.seed(1234)
shape = [5, 4, 3]
Y = smurff.SparseTensor(
pd.DataFrame({
"A": np.random.randint(0, 5, 7),
"B": np.random.randint(0, 4, 7),
"C": np.random.randint(0, 3, 7),
"value": np.random.randn(7)
}), shape)
Ytest = smurff.SparseTensor(
pd.DataFrame({
"A": np.random.randint(0, 5, 5),
"B": np.random.randint(0, 4, 5),
"C": np.random.randint(0, 3, 5),
"value": np.random.randn(5)
}), shape)
predictions = smurff.smurff(Y,
Ytest=Ytest,
priors=['normal', 'normal', 'normal'],
num_latent=4,
verbose=False,
burnin=50,
nsamples=50)
def test_bpmf_dense_matrix_sparse_2d_tensor(self):
np.random.seed(1234)
# Generate train dense matrix
train_shape = (5 ,5)
train_sparse_matrix = scipy.sparse.random(5, 5, density=1.0)
train_dense_matrix = train_sparse_matrix.todense()
# Generate test sparse matrix
test_shape = (5, 5)
test_rows = np.random.randint(0, 5, 5)
test_cols = np.random.randint(0, 4, 5)
test_vals = np.random.randn(5)
test_sparse_matrix = scipy.sparse.coo_matrix((test_vals, (test_rows, test_cols)), test_shape)
# Create train and test sparse tensors
train_sparse_tensor = smurff.SparseTensor(pd.DataFrame({
'0': train_sparse_matrix.row,
'1': train_sparse_matrix.col,
'v': train_sparse_matrix.data
}), train_shape)
test_sparse_tensor = smurff.SparseTensor(pd.DataFrame({
'0': test_sparse_matrix.row,
'1': test_sparse_matrix.col,
'v': test_sparse_matrix.data
}), train_shape)
# Run SMURFF
sparse_matrix_predictions = smurff.smurff(train_dense_matrix,
Ytest=test_sparse_matrix,
priors=['normal', 'normal'],
num_latent=4,
verbose=False,
burnin=50,
nsamples=50,
seed=1234)
sparse_tensor_predictions = smurff.smurff(train_sparse_tensor,
Ytest=test_sparse_tensor,
priors=['normal', 'normal'],
num_latent=4,
verbose=False,
burnin=50,
nsamples=50,
seed=1234)
# Transfrom SMURFF results to dictionary of coords and predicted values
sparse_matrix_results_dict = collections.OrderedDict((p.coords, p.pred_1sample) for p in sparse_matrix_predictions)
sparse_tensor_results_dict = collections.OrderedDict((p.coords, p.pred_1sample) for p in sparse_tensor_predictions)
self.assertEqual(len(sparse_matrix_results_dict), len(sparse_tensor_results_dict))
self.assertEqual(sparse_tensor_results_dict.keys(), sparse_tensor_results_dict.keys())
for coords, matrix_pred_1sample in sparse_matrix_results_dict.items():
tensor_pred_1sample = sparse_tensor_results_dict[coords]
self.assertAlmostEqual(matrix_pred_1sample, tensor_pred_1sample)
def test_bpmf_dense_matrix_dense_2d_tensor(self):
np.random.seed(1234)
# Generate train matrix rows, cols and vals
train_shape = (5, 4)
sparse_random = sp.random(5, 4, density=1.0)
train_dense_matrix = sparse_random.todense()
_, test_sparse_matrix = smurff.make_train_test(sparse_random, 0.2)
# Create train and test sparse
train_sparse_matrix = sp.coo_matrix(train_dense_matrix) # acutally dense
test_sparse_matrix = test_sparse_matrix.tocoo()
# Create train and test sparse representations of dense tensors
train_sparse_tensor = smurff.SparseTensor(pd.DataFrame({
'0': train_sparse_matrix.row,
'1': train_sparse_matrix.col,
'v': train_sparse_matrix.data
}), train_shape)
test_sparse_tensor = smurff.SparseTensor(pd.DataFrame({
'0': test_sparse_matrix.row,
'1': test_sparse_matrix.col,
'v': test_sparse_matrix.data
}), train_shape)
# Run SMURFF
sparse_matrix_predictions = smurff.bpmf(train_dense_matrix,
Ytest=test_sparse_matrix,
num_latent=4,
num_threads=1,
verbose=verbose,
burnin=50,
nsamples=50,
seed=1234)
sparse_tensor_predictions = smurff.bpmf(train_sparse_tensor,
Ytest=test_sparse_tensor,
num_latent=4,
num_threads=1,
verbose=verbose,
burnin=50,
nsamples=50,
seed=1234)
# Sort and compare coords and predicted values
sparse_matrix_predictions.sort()
sparse_tensor_predictions.sort()
self.assertEqual(len(sparse_matrix_predictions), len(sparse_tensor_predictions))
for m, t in zip(sparse_matrix_predictions, sparse_tensor_predictions):
self.assertEqual(m.coords, t.coords)
self.assertAlmostEqual(m.pred_1sample, t.pred_1sample)
def train_test(self):
np.random.seed(1234)
train_df = pd.DataFrame({
"A": np.random.randint(0, 15, 7),
"B": np.random.randint(0, 4, 7),
"C": np.random.randint(0, 3, 7),
"value": np.random.randn(7)
})
test_df = pd.DataFrame({
"A": np.random.randint(0, 15, 5),
"B": np.random.randint(0, 4, 5),
"C": np.random.randint(0, 3, 5),
"value": np.random.randn(5)
})
shape = [15, 4, 3]
Ytrain = smurff.SparseTensor(train_df, shape=shape)
Ytest = smurff.SparseTensor(test_df, shape=shape)
return Ytrain, Ytest
def test_bpmf_tensor(self):
np.random.seed(1234)
Y = smurff.SparseTensor(pd.DataFrame({
"A": np.random.randint(0, 5, 7),
"B": np.random.randint(0, 4, 7),
"C": np.random.randint(0, 3, 7),
"value": np.random.randn(7)
}))
Ytest = smurff.SparseTensor(pd.DataFrame({
"A": np.random.randint(0, 5, 5),
"B": np.random.randint(0, 4, 5),
"C": np.random.randint(0, 3, 5),
"value": np.random.randn(5)
}))
predictions = smurff.bpmf(Y,
Ytest=Ytest,
num_latent=4,
verbose=verbose,
burnin=50,
nsamples=50)
def test_macau_tensor_empty(self):
A = np.random.randn(30, 2)
B = np.random.randn(4, 2)
C = np.random.randn(2, 2)
idx = list( itertools.product(np.arange(A.shape[0]), np.arange(B.shape[0]), np.arange(C.shape[0])) )
df = pd.DataFrame( np.asarray(idx), columns=["A", "B", "C"])
df["value"] = np.array([ np.sum(A[i[0], :] * B[i[1], :] * C[i[2], :]) for i in idx ])
Acoo = scipy.sparse.coo_matrix(A)
predictions = smurff.smurff(smurff.SparseTensor(df),
priors=['normal', 'normal', 'normal'],
num_latent=2,
burnin=5,
nsamples=5,
verbose=False)
self.assertFalse(predictions)
def test_bpmf_sparse_matrix_sparse_2d_tensor(self):
np.random.seed(1234)
# Generate train matrix rows, cols and vals
train_shape = (5, 5)
train_rows = np.random.randint(0, 5, 7)
train_cols = np.random.randint(0, 4, 7)
train_vals = np.random.randn(7)
# Generate test matrix rows, cols and vals
test_shape = (5, 5)
test_rows = np.random.randint(0, 5, 5)
test_cols = np.random.randint(0, 4, 5)
test_vals = np.random.randn(5)
# Create train and test sparse matrices
train_sparse_matrix = scipy.sparse.coo_matrix(
(train_vals, (train_rows, train_cols)), train_shape)
test_sparse_matrix = scipy.sparse.coo_matrix(
(test_vals, (test_rows, test_cols)), test_shape)
# Force NNZ recalculation to remove duplicate coordinates because of random generation
train_sparse_matrix.count_nonzero()
test_sparse_matrix.count_nonzero()
# Create train and test sparse tensors
train_sparse_tensor = smurff.SparseTensor(
pd.DataFrame({
'0': train_sparse_matrix.row,
'1': train_sparse_matrix.col,
'v': train_sparse_matrix.data
}), train_shape)
test_sparse_tensor = smurff.SparseTensor(
pd.DataFrame({
'0': test_sparse_matrix.row,
'1': test_sparse_matrix.col,
'v': test_sparse_matrix.data
}), train_shape)
# Run SMURFF
sparse_matrix_predictions = smurff.bpmf(train_sparse_matrix,
Ytest=test_sparse_matrix,
num_latent=4,
num_threads=1,
verbose=verbose,
burnin=50,
nsamples=50,
seed=1234)
sparse_tensor_predictions = smurff.bpmf(train_sparse_tensor,
Ytest=test_sparse_tensor,
num_latent=4,
num_threads=1,
verbose=verbose,
burnin=50,
nsamples=50,
seed=1234)
# Transfrom SMURFF results to dictionary of coords and predicted values
sparse_matrix_predictions_dict = collections.OrderedDict(
(p.coords, p.pred_1sample) for p in sparse_matrix_predictions)
sparse_tensor_predictions_dict = collections.OrderedDict(
(p.coords, p.pred_1sample) for p in sparse_tensor_predictions)
self.assertEqual(len(sparse_matrix_predictions_dict),
len(sparse_tensor_predictions_dict))
self.assertEqual(sparse_tensor_predictions_dict.keys(),
sparse_tensor_predictions_dict.keys())
for coords, matrix_pred_1sample in sparse_matrix_predictions_dict.items(
):
tensor_pred_1sample = sparse_tensor_predictions_dict[coords]
self.assertAlmostEqual(matrix_pred_1sample, tensor_pred_1sample)
