ProxFunResult

ProxFunResult is the result container returned by the estimator-style ProxFun API.

It keeps link prediction scores in a pandas DataFrame while exposing convenience methods for inspection, ranking, and plotting.

Basic usage

import networkx as nx

from relationalstats.linkprediction import ProxFun

G = nx.path_graph(3)

result = ProxFun(metrics=["jaccard", "adamic_adar"]).fit(G)

df = result.to_dataframe()
top_pairs = result.top_k(k=10, metric="jaccard")

Stored attributes

Attribute	Description
scores_	DataFrame with source, target, and score columns.
metrics_	Logical metric names requested by the estimator.
directed_	Whether directed graph semantics were used.

Convenience properties

result.pair_columns
result.score_columns
result.is_empty

DataFrame helpers

result.to_dataframe()
result.pairs()
result.metric_frame(["jaccard", "adamic_adar"])

Ranking helpers

By default, top_k assumes higher values are better:

result.top_k(k=20, metric="jaccard")

For distance-like metrics such as shortest_path, use ascending ranking:

result.top_k(k=20, metric="shortest_path", ascending=True)

Plotting

Plotting is optional and imports matplotlib lazily:

result.plot_score_distribution("jaccard")

Notes

metrics_ stores the logical metric request. Concrete DataFrame columns may differ for composite metrics. For example, the degree metric produces:

degree_source
degree_target

Use result.score_columns to inspect the actual score columns available for ranking or export.