()
| 2 | |
| 3 | |
| 4 | def bootstrap(): |
| 5 | import pathlib |
| 6 | from datetime import datetime, timedelta |
| 7 | |
| 8 | import numpy as np |
| 9 | import pandas as pd |
| 10 | |
| 11 | from feast.driver_test_data import create_driver_hourly_stats_df |
| 12 | |
| 13 | repo_path = pathlib.Path(__file__).parent.absolute() / "feature_repo" |
| 14 | project_name = pathlib.Path(__file__).parent.absolute().name |
| 15 | data_path = repo_path / "data" |
| 16 | data_path.mkdir(exist_ok=True) |
| 17 | |
| 18 | end_date = datetime.now().replace(microsecond=0, second=0, minute=0) |
| 19 | start_date = end_date - timedelta(days=15) |
| 20 | |
| 21 | # Generate driver data using Feast's built-in test data generator |
| 22 | driver_entities = [1001, 1002, 1003, 1004, 1005] |
| 23 | driver_df = create_driver_hourly_stats_df(driver_entities, start_date, end_date) |
| 24 | |
| 25 | if driver_df["event_timestamp"].dt.tz is None: |
| 26 | driver_df["event_timestamp"] = driver_df["event_timestamp"].dt.tz_localize( |
| 27 | "UTC" |
| 28 | ) |
| 29 | if "created" in driver_df.columns and driver_df["created"].dt.tz is None: |
| 30 | driver_df["created"] = driver_df["created"].dt.tz_localize("UTC") |
| 31 | |
| 32 | driver_stats_path = data_path / "driver_stats.parquet" |
| 33 | driver_df.to_parquet(path=str(driver_stats_path), allow_truncated_timestamps=True) |
| 34 | |
| 35 | # Generate customer data to demonstrate Ray's multi-source capabilities |
| 36 | customer_entities = [2001, 2002, 2003, 2004, 2005] |
| 37 | |
| 38 | # Create customer daily profile data |
| 39 | customer_data = [] |
| 40 | for customer_id in customer_entities: |
| 41 | for i, single_date in enumerate( |
| 42 | pd.date_range(start_date, end_date, freq="D", tz="UTC") |
| 43 | ): |
| 44 | stable_timestamp = single_date.replace( |
| 45 | hour=12, minute=0, second=0, microsecond=0 |
| 46 | ) |
| 47 | customer_data.append( |
| 48 | { |
| 49 | "customer_id": customer_id, |
| 50 | "event_timestamp": stable_timestamp, |
| 51 | "created": stable_timestamp + timedelta(minutes=10), |
| 52 | "current_balance": np.random.uniform(10.0, 1000.0), |
| 53 | "avg_passenger_count": np.random.uniform(1.0, 4.0), |
| 54 | "lifetime_trip_count": np.random.randint(50, 500), |
| 55 | } |
| 56 | ) |
| 57 | |
| 58 | customer_df = pd.DataFrame(customer_data) |
| 59 | |
| 60 | if customer_df["event_timestamp"].dt.tz is None: |
| 61 | customer_df["event_timestamp"] = customer_df["event_timestamp"].dt.tz_localize( |
no test coverage detected