import os
import os.path as osp
from typing import Callable, List, Optional
import mlx.core as mx
import numpy as np
from mlx_graphs.data import HeteroGraphData
from mlx_graphs.datasets.dataset import Dataset
from mlx_graphs.datasets.utils import download, extract_archive
[docs]
class MovieLens100K(Dataset):
"""
The MovieLens 100K heterogeneous rating dataset, assembled by GroupLens
Research from the `MovieLens web site <https://movielens.org>`__,
consisting of movies (1,682 nodes) and users (943 nodes) with 100K
ratings between them.
User ratings for movies are available as ground truth labels.
Features of users and movies are encoded according to the `"Inductive
Matrix Completion Based on Graph Neural Networks"
<https://arxiv.org/abs/1904.12058>`__ paper.
Args:
base_dir (str): Directory where to store dataset files.
transform (callable, optional): A function/transform that takes in an
:obj:`HeteroGraphData` object and returns a
transformed version. The data object will be transformed before
every access. (default: :obj:`None`)
pre_transform (callable, optional): A function/transform that takes in
an `HeteroGraphData` object and returns a
transformed version. The data object will be transformed before
being saved to disk. (default: :obj:`None`)
"""
file_id = "1ggYlYf2_kTyi7oF9g07oTNn3VDhjl7so"
def __init__(
self,
base_dir: str,
transform: Optional[Callable] = None,
pre_transform: Optional[Callable] = None,
):
super().__init__(
name="MovieLens100K",
base_dir=base_dir,
transform=transform,
pre_transform=pre_transform,
)
@property
def raw_path(self) -> str:
return f"{super(self.__class__, self).raw_path}"
@property
def raw_file_names(self) -> List[str]:
return ["u.item", "u.user", "u1.base", "u1.test"]
def download(self):
url = "https://files.grouplens.org/datasets/movielens/ml-100k.zip"
self.data_path = "ml-100k"
path = download(url, self.raw_path)
extract_archive(path, self.raw_path)
os.remove(path)
def process(self):
data = HeteroGraphData(edge_index_dict={})
self.raw_paths = [
osp.join(f"{self.raw_path}/{self.data_path}", self.raw_file_names[i])
for i in range(len(self.raw_file_names))
]
df = np.loadtxt(
self.raw_paths[0], delimiter="|", dtype=str, encoding="ISO-8859-1"
)
movie_mapping = {idx: i for i, idx in enumerate(df[:, 0].astype(int))}
x = df[:, 6:].astype(float)
data.node_features_dict = {"movie": mx.array(x, dtype=mx.float32)}
df = np.loadtxt(
self.raw_paths[1], delimiter="|", dtype=str, encoding="ISO-8859-1"
)
user_mapping = {idx: i for i, idx in enumerate(df[:, 0].astype(int))}
age = df[:, 1].astype(float) / df[:, 1].astype(float).max()
age = mx.array(age, mx.float32).reshape(-1, 1)
gender = np.eye(np.unique(df[:, 2]).size)[
np.unique(df[:, 2], return_inverse=True)[1]
]
gender = mx.array(gender, mx.float32)
occupation = np.eye(np.unique(df[:, 3]).size)[
np.unique(df[:, 3], return_inverse=True)[1]
]
occupation = mx.array(occupation, mx.float32)
data.node_features_dict["user"] = mx.concatenate(
[age, gender, occupation], axis=-1
)
df = np.loadtxt(self.raw_paths[2], delimiter="\t")
src = [user_mapping[idx] for idx in df[:, 0]]
dst = [movie_mapping[idx] for idx in df[:, 1]]
edge_index = mx.array([src, dst], mx.int64)
data.edge_index_dict[("user", "rates", "movie")] = edge_index
rating = mx.array(df[:, 2], dtype=mx.int64)
data.user_rates_movie_rating = rating
time = mx.array(df[:, 3], dtype=mx.int64)
data.user_rates_movie_time = time
data.edge_index_dict[("movie", "rated_by", "user")] = mx.array(
[dst, src], mx.int64
)
data.movie_rated_by_user_time = time
df = np.loadtxt(self.raw_paths[3], delimiter="\t")
src = [user_mapping[idx] for idx in df[:, 0]]
dst = [movie_mapping[idx] for idx in df[:, 1]]
edge_label_index = mx.array([src, dst], mx.int64)
data.edge_labels_index_dict = {}
data.edge_labels_index_dict[("user", "rates", "movie")] = edge_label_index
edge_label = mx.array(df[:, 2], mx.float32)
data.edge_labels_dict = {}
data.edge_labels_dict[("user", "rates", "movie")] = edge_label
if self.pre_transform is not None:
data = self.pre_transform(data)
self.graphs = [data]
