Network metrics

TL;DR

MENoBiS computes core directed non-binary statistics in Rust. NetworkX and rustworkx are optional user choices: install them in your own environment and convert an EdgeTable at the boundary of your application.

Built-in MENoBiS metrics

API	Runtime	Purpose
directed_strengths(edges)	Rust via Python	out/in event totals
directed_degrees(edges)	Rust via Python	out/in binary degrees
compute_all_stats(edges)	Rust via Python	strengths, degrees, Y2, nearest-neighbour stats
occupation_distribution(edges)	Rust via Python	occupation-count histogram
clustering_coefficient(edges)	Rust via Python	binary-support clustering
occupation_clustering_coefficient(edges)	Rust via Python	occupation-based clustering

from menobis.analysis import compute_all_stats
from menobis.data.io import read_edges

edges = read_edges("edges.csv")
stats = compute_all_stats(edges)
print(stats.strength_out)

Optional NetworkX recipe

NetworkX is useful for exploratory Python-only metrics. Keep this adapter in your project so MENoBiS stays independent of NetworkX release cycles.

python -m pip install networkx

import networkx as nx

from menobis.data import read_edges

edges = read_edges("edges.csv")
graph = nx.DiGraph()
graph.add_weighted_edges_from(
    zip(edges.source.tolist(), edges.target.tolist(), edges.occ_num.tolist())
)
centrality = nx.pagerank(graph, weight="occ_num")

Optional rustworkx recipe

import rustworkx as rx

from menobis.data.io import read_edges

edges = read_edges("edges.csv")
graph = rx.PyDiGraph(multigraph=True)
if len(edges) > 0:
    graph.add_nodes_from(range(int(max(edges.source.max(), edges.target.max())) + 1))
for source, target, occ_num in zip(edges.source, edges.target, edges.occ_num, strict=True):
    graph.add_edge(int(source), int(target), int(occ_num))

Dependency policy

External graph libraries remain downstream adapters unless MENoBiS adopts a metric as supported core functionality with Rust tests and docs.