softgraph 0.2.0

Graph ML in one call: softgraph(task=..., dataset=..., ...)

softgraph (one-call API)

Import once, call once, get results.

import softgraph

# Node classification on SBM in one call
run = softgraph.softgraph(
    task="node_classification",
    dataset="sbm",
    n=600, k=4, p_in=0.09, p_out=0.02,
    features="degree+pagerank+cluster",
    model="logreg",
    test_size=0.2,
    random_state=42,
    plot=True
)

print(run["metrics"])

Supported tasks

• node_classification (dataset: sbm)
• edge_classification (dataset: signed_sbm)
• temporal_link_prediction (dataset: temporal_contacts)

All run objects include:

{
  "task": "...",
  "metrics": {...},
  "artifacts": {...}  # e.g., figures
}