def _get_dataset():
path = _transport.get_data(
'https://github.com/zygmuntz/goodbooks-10k/'
'releases/download/v1.0/goodbooks-10k.hdf5', 'goodbooks',
'goodbooks.hdf5')
with h5py.File(path, 'r') as data:
return (data['ratings'][:, 0], data['ratings'][:, 1],
data['ratings'][:, 2].astype(np.float32),
np.arange(len(data['ratings']), dtype=np.int32))
def _get_movielens(dataset):
extension = '.hdf5'
path = _transport.get_data(
os.path.join(URL_PREFIX, VERSION, dataset + extension), 'movielens',
'movielens_{}{}'.format(dataset, extension))
with h5py.File(path, 'r') as data:
return (data['/user_id'][:], data['/item_id'][:], data['/rating'][:],
data['/timestamp'][:])
def _download_amazon():
extension = '.hdf5'
url = ('https://github.com/maciejkula/recommender_datasets/'
'releases/download')
version = '0.1.0'
path = _transport.get_data(
'/'.join((url, version, 'amazon_co_purchasing' + extension)), 'amazon',
'amazon_co_purchasing{}'.format(extension))
with h5py.File(path, 'r') as data:
return (data['/user_id'][:], data['/item_id'][:], data['/rating'][:],
data['/timestamp'][:], data['/features_item_id'][:],
data['/features_feature_id'][:])
def _get_movielens(dataset):
extension = '.hdf5'
path = _transport.get_data(os.path.join(URL_PREFIX,
VERSION,
dataset + extension),
'movielens',
'movielens_{}{}'.format(dataset,
extension))
with h5py.File(path, 'r') as data:
return (data['/user_id'][:],
data['/item_id'][:],
data['/rating'][:],
data['/timestamp'][:])