API: features

`StructuralNodeFeatures`

NEExT.features.StructuralNodeFeatures computes the built-in structural features. The framework’s compute_node_features wraps it.

StructuralNodeFeatures

StructuralNodeFeatures(
  graph_collection,
  feature_list,                 # e.g. ["all"] or ["page_rank", "lsme"]
  feature_vector_length=3,
  normalize_features=True,
  show_progress=True,
  n_jobs=1,
  suffix="",
  parallel_backend="loky",      # "loky" | "threading"
  profile_features=False,
  joblib_kwargs=None,
)

# methods
compute() -> Features
register_metric(name, func) -> None   # register a custom feature function

Built-in feature names: page_rank, degree_centrality, closeness_centrality, betweenness_centrality, eigenvector_centrality, clustering_coefficient, local_efficiency, lsme, load_centrality, basic_expansion, betastar. See Structural features.

`Features`

NEExT.features.Features — a container around the feature DataFrame.

Features

Features(features_df, feature_columns)

# attributes
features.features_df         # DataFrame: node_id, graph_id, <feature columns>
features.feature_columns     # List[str]

# methods
normalize(type="StandardScaler")   # "StandardScaler" | "MinMaxScaler" | "RobustScaler"
features_a + features_b            # merge on (node_id, graph_id) -> Features