Home > Notebooks > Bronze Layer¶

🥉 Bronze Layer Notebooks¶

Purpose: Raw data ingestion from casino source systems with minimal transformation. Append-only landing zone preserving original data fidelity.

The Bronze layer captures raw data from slot machines, player systems, financial transactions, table games, and security systems. Data arrives in its original format with only metadata enrichment (ingestion timestamp, source system, batch ID).

📚 Notebook Inventory¶

⚡ Execution Order¶

graph LR
    A[01_bronze_slot_telemetry] --> B[02_bronze_player_profile]
    B --> C[03_bronze_financial_txn]
    C --> D[04_bronze_compliance]
    D --> E[05_bronze_table_games]
    E --> F[06_bronze_security_events]

    style A fill:#cd7f32
    style B fill:#cd7f32
    style C fill:#cd7f32
    style D fill:#cd7f32
    style E fill:#cd7f32
    style F fill:#cd7f32

Note: Notebooks can run in parallel if source systems are independent. Sequential execution ensures consistent timestamps for lineage tracking.

🔧 Key Transformations¶

Common Bronze Patterns¶

# Standard Bronze ingestion pattern
df = (spark.read
    .format("json")
    .load(source_path)
    .withColumn("_ingested_at", current_timestamp())
    .withColumn("_source_system", lit("slot_machines"))
    .withColumn("_batch_id", lit(batch_id))
)

df.write.format("delta").mode("append").saveAsTable("bronze.slot_telemetry")

📋 Dependencies¶

Pre-requisites Checklist¶

• Fabric workspace created
• Bronze lakehouse provisioned
• Source system credentials configured in Key Vault
• Sample data uploaded to Files/raw/ folder

✅ Validation Steps¶

Data Quality Checks¶

Validation Commands¶

# Verify table exists and has data
display(spark.sql("SELECT COUNT(*) as record_count FROM bronze.slot_telemetry"))

# Check for recent ingestion
display(spark.sql("""
    SELECT 
        DATE(_ingested_at) as ingestion_date,
        COUNT(*) as records
    FROM bronze.slot_telemetry
    GROUP BY DATE(_ingested_at)
    ORDER BY ingestion_date DESC
    LIMIT 7
"""))

# Validate no null keys
assert spark.sql("SELECT COUNT(*) FROM bronze.slot_telemetry WHERE machine_id IS NULL").first()[0] == 0

📖 Related Resources¶

• Tutorial: Day 1: Bronze Ingestion Tutorial
• Architecture: Medallion Architecture Overview
• Data Generation: Bronze Data Generators
• Next Layer: Silver Layer Notebooks

🎰 Casino Domain Context¶

Slot Telemetry Data Points¶

Regulatory Considerations¶

• NIGC MICS: Minimum Internal Control Standards require complete audit trails
• W-2G Threshold: Track wins ≥ $1,200 for IRS reporting
• SAR Patterns: Preserve raw data for suspicious activity detection
• Data Retention: 5+ years for compliance requirements

Next Steps: After Bronze layer completes successfully, proceed to Silver Layer for data cleansing and validation.