Module ml 

Machine learning utilities for graph neural networks.

This module provides comprehensive feature engineering for graph-based ML:

• Structural features (degree, clustering, etc.)
• Temporal sequence features (velocity, burst, trend)
• Motif detection (cycles, stars, back-and-forth)
• Relationship features (counterparty concentration, risk)
• Entity group detection and aggregation

Re-exports

pub use aggregation::aggregate_all_groups;

pub use aggregation::aggregate_features;

pub use aggregation::aggregate_node_features;

pub use aggregation::aggregate_values;

pub use aggregation::aggregate_weighted;

pub use aggregation::AggregatedFeatures;

pub use aggregation::AggregationType;

pub use aggregation::MultiFeatureAggregation;

pub use entity_groups::detect_entity_groups;

pub use entity_groups::EntityGroup;

pub use entity_groups::GroupDetectionAlgorithm;

pub use entity_groups::GroupDetectionConfig;

pub use entity_groups::GroupDetectionResult;

pub use entity_groups::GroupType;

pub use motifs::compute_motif_features;

pub use motifs::detect_motifs;

pub use motifs::find_back_and_forth;

pub use motifs::find_circular_flows;

pub use motifs::find_star_patterns;

pub use motifs::CircularFlow;

pub use motifs::GraphMotif;

pub use motifs::MotifConfig;

pub use motifs::MotifDetectionResult;

pub use motifs::MotifInstance;

pub use relationship_features::compute_all_combined_features;

pub use relationship_features::compute_all_counterparty_risk;

pub use relationship_features::compute_all_relationship_features;

pub use relationship_features::compute_counterparty_risk;

pub use relationship_features::compute_relationship_features;

pub use relationship_features::CombinedRelationshipFeatures;

pub use relationship_features::CounterpartyRisk;

pub use relationship_features::RelationshipFeatureConfig;

pub use relationship_features::RelationshipFeatures;

pub use temporal::compute_all_temporal_features;

pub use temporal::compute_temporal_sequence_features;

pub use temporal::TemporalConfig;

pub use temporal::TemporalFeatures;

pub use temporal::TemporalIndex;

pub use temporal::WindowFeatures;

Modules

aggregation

Feature aggregation for entity groups.

entity_groups

Entity group detection for collective fraud analysis.

motifs

Subgraph and motif detection for fraud pattern identification.

relationship_features

Entity relationship feature computation for fraud detection.

temporal

Temporal sequence feature computation for graph nodes.

Structs

DataSplit

Result of a data split.

DataSplitter

Data splitter for graph data.

FeatureNormalizer

Feature normalizer for graph features.

SplitConfig

Configuration for data splitting.

Enums

NormalizationMethod

Feature normalization method.

SplitStrategy

Strategy for splitting data.

Functions

compute_amount_features

Computes amount-based features.

compute_benford_features

Computes Benford’s law features for an amount.

compute_edge_direction_features

Computes edge direction features for directed graphs.

compute_structural_features

Computes structural features for nodes.

compute_temporal_features

Computes temporal features for edges.

label_encode

Label encodes a categorical value.

one_hot_encode

One-hot encodes a categorical value.

positional_encoding

Creates positional encoding for graph nodes (similar to transformer positional encoding).

sample_negative_edges

Creates negative edge samples for link prediction.