# feast init -t ray
Ray template initialized with sample data:
  - Driver stats: /feast-projects/sacred_rodent/feature_repo/data/driver_stats.parquet
  - Customer profiles: /feast-projects/sacred_rodent/feature_repo/data/customer_daily_profile.parquet
  - Ray storage will be created at: /feast-projects/sacred_rodent/feature_repo/data/ray_storage

To get started:
  1. cd sacred_rodent/feature_repo
  2. feast apply
  3. python test_workflow.py

Creating a new Feast repository in /feast-projects/sacred_rodent.

# python test_workflow.py
============================================================
🚀 Ray Offline Store & Compute Engine Demo
============================================================

1. Initializing Feast with Ray configuration...
   ✓ Offline store: ray
   ✓ Compute engine: ray.engine

2. Creating entity DataFrame for historical feature retrieval...
   ✓ Created entity DataFrame with 3 rows
   ✓ Time range: 2025-08-15 12:00:00 to 2025-08-17 12:00:00

3. Retrieving historical features with Ray compute engine...
   (This demonstrates distributed point-in-time joins)
   ✓ Retrieved 3 historical feature rows
   ✓ Features: ['driver_id', 'customer_id', 'event_timestamp', 'conv_rate', 'acc_rate', 'avg_daily_trips', 'current_balance', 'avg_passenger_count', 'lifetime_trip_count']

   Sample historical features:
 driver_id  customer_id           event_timestamp  conv_rate  acc_rate  avg_daily_trips  current_balance  avg_passenger_count  lifetime_trip_count
      1001         2001 2025-08-16 12:00:00+00:00   0.253998  0.016639               76       738.030300             1.841441                  387
      1002         2002 2025-08-17 00:00:00+00:00   0.734615  0.839181               79        83.531244             2.651579                  342
      1003         2003 2025-08-17 06:00:00+00:00   0.768238  0.294502              968        39.029157             3.438895                  215

4. Testing on-demand feature transformations...
   ✓ Retrieved 1 rows with on-demand transformations

   Sample with on-demand features:
 driver_id  conv_rate  acc_rate  conv_rate_plus_acc_rate
      1001   0.253998  0.016639                 0.270637

5. Materializing features to online store...
   Attempting materialization up to 2025-08-17 12:00:00
Materializing 2 feature views to 2025-08-17 12:00:00+00:00 into the sqlite online store.

driver_hourly_stats from 2025-08-10 06:31:21+00:00 to 2025-08-17 12:00:00+00:00:
customer_daily_profile from 2025-08-10 06:31:26+00:00 to 2025-08-17 12:00:00+00:00:
   ✓ Ray compute engine materialization successful!

6. Testing online feature serving...
   ✓ Retrieved 2 online feature rows

   Sample online features:
 driver_id  customer_id  conv_rate  acc_rate  current_balance
      1001         2001   0.542408  0.615162        55.650719
      1002         2002   0.326722  0.137612       440.106567

============================================================
🎉 Ray Demo Complete!
============================================================
If you want to explore Feast with your existing ray cluster, you can configure ray_address to feature_store.yaml:
    offline_store:
      ray_address: "127.0.0.1:10001"
    batch_engine:
      ray_address: "127.0.0.1:10001"