rel-amazon: item churn (Classification)

This example now uses the official relbench dataset loader and materializes rel-amazon tables locally before running GraphReduce. It no longer downloads parquet files from the deprecated S3 example bucket.

from relbench_amazon_item_churn import run_rel_amazon_item_churn

df, auc, n_features, materialized, target = run_rel_amazon_item_churn()
print(materialized)
print(target, auc, n_features, len(df))

Current implementation:

• examples/relbench_amazon_item_churn.py
• examples/relbench_amazon_common.py

Current RelBench-aligned dates:

• validation cut date: 2015-10-01
• holdout cut date: 2016-01-01
• lookback start: 2008-01-01

Latest Tested Result

Latest local run from tests/data/relbench/run_reports/relbench_results.json:

• test ROC AUC: 0.8249
• test average precision: 0.7012
• feature count: 31
• rows: train 179118, validation 177689, test 166842
• runtime: 93.8s (2026-04-09T21:49:24Z)

Interactive Runner

Save API URL Run rel-amazon item churn

Idle