torch_geometric.datasets.MixHopSyntheticDataset

class MixHopSyntheticDataset(root: str, homophily: float, transform: Optional[Callable] = None, pre_transform: Optional[Callable] = None, force_reload: bool = False)[source]

Bases: InMemoryDataset

The MixHop synthetic dataset from the “MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing” paper, containing 10 graphs, each with varying degree of homophily (ranging from 0.0 to 0.9). All graphs have 5,000 nodes, where each node corresponds to 1 out of 10 classes. The feature values of the nodes are sampled from a 2D Gaussian distribution, which are distinct for each class.

Parameters:

  • root (str) – Root directory where the dataset should be saved.

  • homophily (float) – The degree of homophily (one of 0.0, 0.1, …, 0.9).

  • transform (callable, optional) – A function/transform that takes in an torch_geometric.data.Data object and returns a transformed version. The data object will be transformed before every access. (default: None)

  • pre_transform (callable, optional) – A function/transform that takes in an torch_geometric.data.Data object and returns a transformed version. The data object will be transformed before being saved to disk. (default: None)

  • force_reload (bool, optional) – Whether to re-process the dataset. (default: False)