# Disney (PyGOD)

**Task**: Outlier movie detection (co-purchase)  
**Size band**: small · **Label type**: anomaly  
**Label column**: `is_outlier` · **Converter**: `converters/convert_pygod.py`

Tiny Amazon co-purchase network of Disney movies (124 nodes) with 28 features and binary outlier labels. Smoke-test sized — useful for pipeline debugging, not for meaningful scores.

## Converted graphs (neext/)

| graph | nodes | edges | classes | feature cols | isolated | class counts |
|---|---|---|---|---|---|---|
| default | 124 | 335 | 2 | 28 | 0 | 0: 118, 1: 6 |

*Conversion notes*: PyGOD .pt (torch pickle, trusted pygod-team/data source) via stub unpickling; edge_index symmetrized; 28 x_* feature columns.

## Source

- [disney.pt.zip](https://raw.githubusercontent.com/pygod-team/data/main/disney.pt.zip) — 8,987 bytes, sha256 `761d111a6ec9d545…`, fetched 2026-07-23

**License**: MIT (pygod-team/data)  
**Citation**: Liu et al. BOND. NeurIPS 2022 D&B.

- https://github.com/pygod-team/data

## Caveats

- Only 124 nodes / ~6 outliers; treat results as sanity checks only.

---
*Generated by converters/make_cards.py; stats from metadata.json. Raw files: `source/`. NEExT tables: `neext/`.*