def test_sparse_to_df(test_specs, python_dataset):
# initialize the splitter
header = {
"col_user": DEFAULT_USER_COL,
"col_item": DEFAULT_ITEM_COL,
"col_rating": DEFAULT_RATING_COL,
}
# instantiate the the affinity matrix
am = AffinityMatrix(DF=python_dataset, **header)
# generate the sparse matrix representation
X, _, _ = am.gen_affinity_matrix()
# use the inverse function to generate a pandas df from a sparse matrix ordered by userID
DF = am.map_back_sparse(X, kind="ratings")
# tests: check that the two dataframes have the same elements in the same positions.
assert (
DF.userID.values.all()
== python_dataset.sort_values(by=["userID"]).userID.values.all()
)
assert (
DF.itemID.values.all()
== python_dataset.sort_values(by=["userID"]).itemID.values.all()
)
assert (
DF.rating.values.all()
== python_dataset.sort_values(by=["userID"]).rating.values.all()
)
def test_sparse_to_df(test_specs, python_dataset):
# initialize the splitter
header = {
"col_user": DEFAULT_USER_COL,
"col_item": DEFAULT_ITEM_COL,
"col_rating": DEFAULT_RATING_COL,
}
# instantiate the the affinity matrix
am = AffinityMatrix(DF=python_dataset, **header)
# generate the sparse matrix representation
X = am.gen_affinity_matrix()
# use the inverse function to generate a pandas df from a sparse matrix ordered by userID
DF = am.map_back_sparse(X, kind="ratings")
# tests: check that the two dataframes have the same elements in the same positions.
assert (
DF.userID.values.all()
== python_dataset.sort_values(by=["userID"]).userID.values.all()
)
assert (
DF.itemID.values.all()
== python_dataset.sort_values(by=["userID"]).itemID.values.all()
)
assert (
DF.rating.values.all()
== python_dataset.sort_values(by=["userID"]).rating.values.all()
)
def test_df_to_sparse(test_specs, python_dataset):
# initialize the splitter
header = {
"col_user": DEFAULT_USER_COL,
"col_item": DEFAULT_ITEM_COL,
"col_rating": DEFAULT_RATING_COL,
}
# instantiate the affinity matrix
am = AffinityMatrix(DF=python_dataset, **header)
# obtain the sparse matrix representation of the input dataframe
X, _, _ = am.gen_affinity_matrix()
# check that the generated matrix has the correct dimensions
assert (X.shape[0] == python_dataset.userID.unique().shape[0]) & (
X.shape[1] == python_dataset.itemID.unique().shape[0])
def test_df_to_sparse(test_specs, python_dataset):
# initialize the splitter
header = {
"col_user": DEFAULT_USER_COL,
"col_item": DEFAULT_ITEM_COL,
"col_rating": DEFAULT_RATING_COL,
}
# instantiate the affinity matrix
am = AffinityMatrix(DF=python_dataset, **header)
# obtain the sparse matrix representation of the input dataframe
X = am.gen_affinity_matrix()
# check that the generated matrix has the correct dimensions
assert (X.shape[0] == python_dataset.userID.unique().shape[0]) & (
X.shape[1] == python_dataset.itemID.unique().shape[0]
)
def RBMtrain():
data = pd.read_csv("SnacksData100.csv")
header = {
"col_user": "******",
"col_item": "Product_Id",
"col_rating": "Ratings",
}
am = AffinityMatrix(DF=data, **header)
X = am.gen_affinity_matrix()
Xtr, Xtst = numpy_stratified_split(X)
model = RBM(hidden_units=600,
training_epoch=30,
minibatch_size=60,
keep_prob=0.9,
with_metrics=True)
model.fit(Xtr, Xtst)
top_k, test_time = model.recommend_k_items(Xtst)
top_k_df = am.map_back_sparse(top_k, kind='prediction')
test_df = am.map_back_sparse(Xtst, kind='ratings')
joblib.dump(top_k_df, 'testdata')
smpl = pd.merge(data, titles, on="MovieID").sample(SMPLS)
smpl['MovieTitle'] = smpl['MovieTitle'].str[:TITLEN]
smpl['Rating'] = pd.to_numeric(smpl['Rating'], downcast='integer')
del smpl['Timestamp'] # Drop the column from printing.
print(smpl.to_string())
header = {
"col_user": "******",
"col_item": "MovieID",
"col_rating": "Rating",
}
# Use a sparse matrix representation rather than a pandas data frame
# for significant performance gain.
am = AffinityMatrix(DF=data, **header)
X = am.gen_affinity_matrix()
# Contstruct the training and test datasets.
Xtr, Xtst = numpy_stratified_split(X)
print('\nTraining matrix size (users, movies) is:', Xtr.shape)
print('Testing matrix size is: ', Xtst.shape)
# Initialize the model class. Note that through random variation we
# can get a much better performing model with seed=1!
model = RBM(
hidden_units=600,
training_epoch=30,