Graph Datasets

A unified package for loading popular homo/hetero-philic graph datasets from various sources (PyG, DGL, OGB, LinkX, etc.) with preprocessing and format conversion support (DGL, PyG).

---

🚀 Installation

Requirements

• python>=3.8
• torch>=1.12
• torch_geometric>=2.0
• dgl>=1.1

Install via pip:

$ python -m pip install graph_datasets

---

💡 Usage

For full documentation, see the Graph Datasets.

from graph_datasets import load_data

graph, label, n_clusters = load_data(
    dataset_name='cora',
    directory='./data',
    source='pyg',
    verbosity=1,
    return_type= "dgl",
    row_normalize= True,
    rm_self_loop= True,
    add_self_loop= False,
    to_simple= True,
)

Tip: If you experience issues downloading datasets from GitHub sources, set up a GitHub proxy in your hosts file as described in GitHub-IP-hosts.

---

📊 Statistics

Pull requests are welcome — help us expand the dataset collection! 🚀

idx	source	dataset	n_nodes	n_feats	n_edges	n_clusters
1	pyg	cora	2,708	1,433	10,556	7
2	pyg	citeseer	3,327	3,703	9,104	6
3	pyg	pubmed	19,717	500	88,648	3
4	pyg	corafull	19,793	8,710	126,842	70
5	pyg	reddit	232,965	602	114,615,892	41
6	pyg	chameleon	2,277	2,325	62,742	5
7	pyg	squirrel	5,201	2,089	396,706	5
8	pyg	actor	7,600	932	53,318	5
9	pyg	cornell	183	1,703	554	5
10	pyg	texas	183	1,703	558	5
11	pyg	wisconsin	251	1,703	900	5
12	pyg	computers	13,752	767	491,722	10
13	pyg	photo	7,650	745	238,162	8
14	pyg	cs	18,333	6,805	163,788	15
15	pyg	physics	34,493	8,415	495,924	5
16	pyg	wikics	11,701	300	431,206	10
17	dgl	cora	2,708	1,433	10,556	7
18	dgl	citeseer	3,327	3,703	9,104	6
19	dgl	pubmed	19,717	500	88,648	3
20	dgl	corafull	19,793	8,710	126,842	70
21	dgl	reddit	232,965	602	114,615,892	41
22	dgl	chameleon	2,277	2,325	62,742	5
23	dgl	squirrel	5,201	2,089	396,706	5
24	dgl	actor	7,600	932	53,318	5
25	dgl	cornell	183	1,703	554	5
26	dgl	texas	183	1,703	558	5
27	dgl	wisconsin	251	1,703	900	5
28	ogb	products	2,449,029	100	123,718,024	47
29	ogb	arxiv	169,343	128	2,315,598	40
30	sdcn	dblp	4,057	334	7,056	4
31	sdcn	acm	3,025	1,870	26,256	3
32	cola	blogcatalog	5,196	8,189	343,486	6
33	cola	flickr	7,575	12,047	479,476	9
34	linkx	snap-patents	2,923,922	269	27,945,090	5
35	linkx	pokec	1,632,803	65	44,603,928	3
36	linkx	genius	421,961	12	1,845,736	2
37	linkx	arxiv-year	169,343	128	2,315,598	5
38	linkx	Penn94	41,554	4,814	2,724,458	3
39	linkx	twitch-gamers	168,114	7	13,595,114	2
40	linkx	wiki	1,925,342	600	485,014,138	6
41	linkx	cornell	183	1,703	554	5
42	linkx	chameleon	2,277	2,325	62,742	5
43	linkx	film	7,600	932	53,318	5
44	linkx	squirrel	5,201	2,089	396,706	5
45	linkx	texas	183	1,703	558	5
46	linkx	wisconsin	251	1,703	900	5
47	linkx	yelp-chi	45,954	32	7,693,958	2
48	linkx	deezer-europe	28,281	31,241	185,504	2
49	linkx	Amherst41	2,235	1,193	181,908	3
50	linkx	Cornell5	18,660	4,735	1,581,554	3
51	linkx	Johns Hopkins55	5,180	2,406	373,172	3
52	linkx	Reed98	962	745	37,624	3
53	critical	roman-empire	22,662	300	65,854	18
54	critical	amazon-ratings	24,492	300	186,100	5
55	critical	minesweeper	10,000	7	78,804	2
56	critical	tolokers	11,758	10	1,038,000	2
57	critical	questions	48,921	301	307,080	2
58	critical	squirrel	2,223	2,089	93,996	5
59	critical	chameleon	890	2,325	17,708	5

---

⚙️ Requirements

Dependencies are listed in:

• requirements.txt
• requirements-dev.txt
• pyproject.toml (see dependencies)

---

🤝 Contributing

Contributions are welcome! Please check the CONTRIBUTING.md for guidelines.

---