split_data.py

"""
Split sensor_data.csv into smaller files by person
"""
import pandas as pd
import os

print("Loading sensor_data.csv...")
sensor_df = pd.read_csv("sensor_data.csv")
print(f"Loaded {len(sensor_df):,} rows")

# Create output directory
DATA_DIR = "sensor_data"
os.makedirs(DATA_DIR, exist_ok=True)

print(f"\nSplitting data by person into '{DATA_DIR}/'...")

# Get unique person IDs
person_ids = sorted(sensor_df["PERSON_ID"].unique())
print(f"Found {len(person_ids)} people")

# Save each person's data
for person_id in person_ids:
    person_df = sensor_df[sensor_df["PERSON_ID"] == person_id]
    person_file = os.path.join(DATA_DIR, f"person_{person_id:02d}.csv")
    person_df.to_csv(person_file, index=False)

    file_size_mb = os.path.getsize(person_file) / (1024 * 1024)
    print(f"  ✓ Person {person_id:2d}: {person_file} ({file_size_mb:.1f} MB, {len(person_df):,} rows)")

print(f"\n✅ Split complete! {len(person_ids)} files created in '{DATA_DIR}/'")