API

Primary Imports¶

from distributed_random_forest import (
    RandomForest,
    DPRandomForest,
    ClientRF,
    DPClientRF,
    FederatedAggregator,
    FederatedRandomForest,
    create_partitions,
)

Local Training¶

`RandomForest`¶

Train a local forest with simple or weighted voting.

`DPRandomForest`¶

Train a local forest with built-in differential privacy noise at the tree level.

Federated Training¶

`ClientRF` and `DPClientRF`¶

Client-scoped wrappers that store:

training metrics
validation metrics
class distributions
sample counts

`FederatedAggregator`¶

Use this when you want direct control over tree selection:

aggregator = FederatedAggregator(
    strategy="top_k_global_balanced_accuracy",
    n_total_trees=24,
)

`FederatedRandomForest`¶

Use this when you want an end-to-end workflow with partitioning, parallel client training, strategy evaluation, and JSON reports.

Utilities¶

Partitioning¶

create_partitions
partition_uniform_random
partition_stratified
partition_dirichlet
partition_label_skew
partition_by_feature
partition_random_with_sizes

Metrics¶

compute_accuracy
compute_weighted_accuracy
compute_balanced_accuracy
compute_f1_score
evaluate_predictions

Aggregation Constants¶

AVAILABLE_STRATEGIES
AggregationSummary

API