def _init_model(self):
if self._model is None:
self._model = FM(n_factors=self._n_factors,
sparse=self._sparse,
n_iter=self._n_iter,
loss=self._loss,
l2=self._l2,
learning_rate=self._learning_rate,
optimizer_func=self._optimizer_func,
batch_size=self._batch_size,
random_state=self._random_state,
use_cuda=self._use_cuda,
device_id=self._device_id,
logger=self._logger,
n_jobs=self._n_jobs,
pin_memory=self._pin_memory,
verbose=self._verbose,
early_stopping=self._early_stopping,
n_iter_no_change=self._n_iter_no_change,
tol=self._tol,
stopping=self._stopping)
elif isinstance(self._model, str):
self._model = FM(model=self._model,
n_factors=self._n_factors,
sparse=self._sparse,
n_iter=self._n_iter,
loss=self._loss,
l2=self._l2,
learning_rate=self._learning_rate,
optimizer_func=self._optimizer_func,
batch_size=self._batch_size,
random_state=self._random_state,
use_cuda=self._use_cuda,
device_id=self._device_id,
logger=self._logger,
n_jobs=self._n_jobs,
pin_memory=self._pin_memory,
verbose=self._verbose,
early_stopping=self._early_stopping,
n_iter_no_change=self._n_iter_no_change,
tol=self._tol,
stopping=self._stopping)
elif not isinstance(self._model, FM):
raise ValueError("Model must be an instance of "
"divmachines.classifiers.FM class")
verbose=VERBOSE,
sparse=SPARSE)
train = pd.read_csv(train_path, header=0)
ground = pd.read_csv(ground_path, header=0)
dataset = pd.concat((train, ground))
users = dataset.user.unique()
item_catalogue = dataset[proj[1:-1]].drop_duplicates().values
values = cartesian2D(users.reshape(-1, 1), item_catalogue)
model0 = FM(n_iter=N_ITER,
model=MODEL_PATH,
n_jobs=N_JOBS,
batch_size=BATCH_SIZE,
n_factors=FACTORS,
learning_rate=LEARNING_RATE,
use_cuda=USE_CUDA,
verbose=VERBOSE,
sparse=SPARSE)
rank = model0.predict(values) \
.reshape(users.shape[0], item_catalogue.shape[0])
for b in tqdm([1.0], desc="sys.div.", leave=False):
table = np.zeros((users.shape[0], TOP + 1), dtype=np.object)
table[:, 0] = users
table[:, 1:] = model.predict(values, top=TOP, b=b, rank=rank)
np.savetxt("./results-b" + str(b), table, fmt="%s")
"Musical | Mystery | Romance | Sci-Fi | Thriller | War | Western "
header = header.replace(" |", "")
header = header.split()
items = pd.read_csv(GENRE_PATH, sep="|", names=header, encoding='iso-8859-2')
proj = ['user', 'item']
proj.extend(header[5:])
proj.append('rating')
train = pd.merge(data, items, on='item', how='inner')[proj].sample(1000)
n_users = np.unique(train[["user"]].values).shape[0]
n_items = np.unique(train[["item"]].values).shape[0]
print("Number of users: %s" % n_users)
print("Number of items: %s" % n_items)
model = FM(verbose=True)
interactions = train.values
x = interactions[:, :-1]
y = interactions[:, -1]
gSearch = GridSearchCV(model,
param_grid={"n_iter": [10], "learning_rate": [1], "l2": [0, 0.1, 0.2]},
cv='kFold',
metrics='mean_square_error',
n_jobs=8,
verbose=10,
return_train_score=True)
gSearch.fit(x, y, fit_params={'dic':
{'users': 0, 'items': 1},
train = train.loc[train['user'].isin(user_take)]
dr = test.groupby('user', as_index=False) \
.apply(lambda g: g.sample(5)).reset_index(0, drop=True)
# Create Ground Truth
ground = test[~test.index.isin(dr.index)].dropna()
train = pd.concat((train, dr))
users = ground.user.unique()
model = FM(n_iter=100,
learning_rate=1e-3,
sparse=False,
batch_size=4096,
n_jobs=8,
optimizer_func=Adam,
use_cuda=True,
verbose=True,
stopping=True,
early_stopping=True)
interactions = train[['user', 'item', 'rating']].values
np.random.shuffle(interactions)
x = interactions[:, :-1]
y = interactions[:, -1]
model.fit(x, y, {'users': 0, 'items': 1}, n_items=n_items, n_users=n_users)
x = cartesian2D(users.reshape(-1, 1), item_cat.reshape(-1, 1))
pred = model.predict(x).reshape(users.shape[0], item_cat.shape[0])
"Musical | Mystery | Romance | Sci-Fi | Thriller | War | Western "
header = header.replace(" |", "")
header = header.split()
items = pd.read_csv(GENRE_PATH, sep="|", names=header, encoding='iso-8859-2')
proj = ['user', 'item']
proj.extend(header[5:])
proj.append('rating')
train = pd.merge(data, items, on='item', how='inner')[proj]
n_users = np.unique(train[["user"]].values).shape[0]
n_items = np.unique(train[["item"]].values).shape[0]
print("Number of users: %s" % n_users)
print("Number of items: %s" % n_items)
model = FM(n_iter=10, learning_rate=1e-1, use_cuda=False)
cv = KFold(folds=10, shuffle=True)
interactions = train.values
x = interactions[:, :-1]
y = interactions[:, -1]
for k, v in cross_validate(model,
x,
y,
cv=cv,
fit_params={
'dic': {
'users': 0,
'items': 1
df = pd.read_csv(PATH, sep=",", header=0)
feats = pd.read_csv(FEATS_PATH, sep=",",
header=0)[['artist_id', 'mode', 'tps_id']]
train = pd.merge(df, feats, on="tps_id")
n_users = df.user_id.unique().shape[0]
n_items = df.tps_id.unique().shape[0]
n_artists = feats.artist_id.unique().shape[0]
interactions = train[['user_id', 'tps_id', 'artist_id', 'mode',
'playcounts']].values
logger = TLogger()
model = FM(n_iter=10,
learning_rate=1e-1,
logger=logger,
n_jobs=4,
use_cuda=False)
x = interactions[:, :-1]
y = interactions[:, -1]
model.fit(x,
y, {
'users': 0,
'items': 1,
'artists': 2
},
n_users=n_users,
n_items=n_items,
lengths={"n_artists": n_artists})