Statistics¶

Whole-graph values. Three of them are eager and return a Python number. describe is lazy and returns a one-row frame.

ur.describe¶

ur.describe(edges, full: bool = False) -> NodeFrame

A lazy one-row summary with the columns n_nodes, n_edges, density, and avg_degree. With full=True, the row also has n_components, which costs a components pass.

The result is not keyed by node. Call collect() on it directly; the shared row methods do not apply.

ur.density¶

ur.density(edges) -> float

The directed edge density: m / (n · (n − 1)), the edges present over the edges possible. Self-loops are excluded from the count of possible edges. Parallel edges count as given. For the simple-graph value, build the frame from the distinct rows first; see Reshaped edge sets.

ur.avg_path_length¶

ur.avg_path_length(edges, sample: float | None = None) -> float

The mean shortest-path length over the ordered pairs that have a path, following out-edges. sample (a fraction in (0, 1]) estimates it from a subset of sources.

ur.diameter¶

ur.diameter(edges, approximate: bool = True) -> int

The longest shortest path over the pairs that have a path, following out-edges. The default is a lower-bound estimate. approximate=False gives the exact value, which costs one BFS per node.