Skip to content

Source: examples/casino-analytics/README.md — this page is rendered live from that file.

Tribal Casino Player Analytics Platform

Examples > Casino Analytics

[!TIP] TL;DR — Casino operations analytics with real-time slot telemetry streaming, player engagement scoring, floor optimization, and Title 31 AML compliance. All data is fully synthetic. Includes Event Hub to ADX streaming and NIGC regulatory reporting.

📋 Table of Contents

A comprehensive casino operations analytics platform built on Azure Cloud Scale Analytics (CSA), providing real-time player engagement scoring, floor optimization, food & beverage analytics, and regulatory compliance reporting. All data in this example is fully synthetic — no real player data is included.

📋 Overview

Tribal casinos represent a $40 billion industry operated by 245 tribes across 29 states. Casino operations generate massive volumes of real-time telemetry: slot machines produce thousands of events per second, table games track hand-by-hand results, player tracking systems monitor loyalty program engagement, and surveillance systems ensure regulatory compliance. This platform demonstrates how Cloud Scale Analytics handles high-velocity streaming data alongside batch analytics for a complete casino intelligence solution.

[!WARNING] ALL DATA IS SYNTHETIC — This example uses entirely synthetic data generated for demonstration purposes. No real player information, financial records, or operational data from any tribal casino is included. The synthetic data generators produce realistic but fictional records for: player profiles, slot machine events, table game hands, F&B transactions, hotel reservations, and loyalty program activity.

✨ Key Features

  • Real-Time Player Engagement: Slot machine event streaming via Event Hub → ADX for live dashboards
  • Floor Layout Optimization: Heat mapping and machine placement analytics
  • F&B Preference Analysis: Dining pattern recognition with comp optimization
  • Loyalty Program Intelligence: Player lifecycle tracking and churn prediction
  • Video Analytics: Azure Video Analyzer + Custom Vision for floor density estimation
  • Regulatory Compliance: NIGC reporting, Title 31 CTR/SAR automated detection
  • Suspicious Activity Detection: ML-based anomaly detection for AML compliance

🗄️ Data Sources (All Synthetic)

Source Type Description Generator
Player Tracking System Synthetic Player profiles, rated play, tier status generate_player_data.py
Slot Machine Telemetry Synthetic/Streaming Spin events, denomination, win/loss, bonus triggers slot_event_simulator.py
Table Game Records Synthetic Buy-ins, hands played, avg bet, win/loss by position generate_table_data.py
F&B POS Transactions Synthetic Restaurant/bar orders, comps, timing generate_fnb_data.py
Hotel PMS Records Synthetic Reservations, room types, revenue, stay patterns generate_hotel_data.py
Loyalty Program Synthetic Points earned/redeemed, tier changes, offers accepted generate_loyalty_data.py
Surveillance Events Synthetic Floor density counts, incident flags generate_surveillance_data.py

🏗️ Architecture Overview

graph TD
    subgraph "Real-Time Sources"
        RT1[Slot Machine<br/>Telemetry]
        RT2[Table Game<br/>Shoe Tracking]
        RT3[Video Analytics<br/>Floor Cameras]
    end

    subgraph "Batch Sources"
        BA1[Player Tracking<br/>System]
        BA2[F&B POS<br/>Transactions]
        BA3[Hotel PMS<br/>Reservations]
        BA4[Loyalty Program<br/>Activity]
    end

    subgraph "Streaming Pipeline"
        EH[Azure Event Hub<br/>casino-events]
        ADX[Azure Data Explorer<br/>Hot Analytics]
        RT_DASH[Real-Time<br/>Floor Dashboard]
        ALERT[Alert Engine<br/>Compliance Triggers]
    end

    subgraph "Video Analytics"
        VA1[Azure Video Analyzer<br/>Camera Feeds]
        CV[Custom Vision<br/>Floor Density Model]
        VA_OUT[Density Heatmap<br/>Updates]
    end

    subgraph "Batch Ingestion"
        I1[ADF Pipeline<br/>Nightly Loads]
    end

    subgraph "Bronze Layer — Raw"
        B1[brz_slot_events]
        B2[brz_table_hands]
        B3[brz_player_profiles]
        B4[brz_fnb_transactions]
        B5[brz_hotel_stays]
        B6[brz_loyalty_activity]
        B7[brz_floor_density]
    end

    subgraph "Silver Layer — Cleansed"
        S1[slv_gaming_activity]
        S2[slv_player_360]
        S3[slv_fnb_patterns]
        S4[slv_hospitality]
        S5[slv_floor_analytics]
    end

    subgraph "Gold Layer — Analytics"
        G1[gld_player_engagement]
        G2[gld_floor_optimization]
        G3[gld_fnb_preferences]
        G4[gld_suspicious_activity]
        G5[gld_regulatory_reports]
        G6[gld_casino_dashboard]
    end

    subgraph "Consumption"
        C1[Casino Floor<br/>Live Dashboard]
        C2[Player Host<br/>Mobile App]
        C3[F&B Management<br/>Console]
        C4[Compliance<br/>Reports]
        C5[Executive<br/>Analytics]
    end

    RT1 --> EH
    RT2 --> EH
    EH --> ADX
    ADX --> RT_DASH
    ADX --> ALERT

    RT3 --> VA1
    VA1 --> CV
    CV --> VA_OUT
    VA_OUT --> EH

    BA1 --> I1
    BA2 --> I1
    BA3 --> I1
    BA4 --> I1

    I1 --> B3
    I1 --> B4
    I1 --> B5
    I1 --> B6
    ADX --> B1
    ADX --> B2
    VA_OUT --> B7

    B1 --> S1
    B2 --> S1
    B3 --> S2
    B4 --> S3
    B5 --> S4
    B6 --> S2
    B7 --> S5

    S1 --> G1
    S2 --> G1
    S1 --> G2
    S5 --> G2
    S3 --> G3
    S1 --> G4
    S2 --> G4
    S1 --> G5
    S2 --> G5
    S1 --> G6
    S2 --> G6
    S3 --> G6
    S4 --> G6

    G1 --> C1
    G1 --> C2
    G3 --> C3
    G4 --> C4
    G5 --> C4
    G6 --> C5

⚡ Real-Time Streaming Architecture

The streaming pipeline is the centerpiece of this example, demonstrating high-velocity event processing from slot machines and table games:

graph LR
    subgraph "Gaming Floor"
        SM1[Slot Bank A<br/>50 Machines]
        SM2[Slot Bank B<br/>50 Machines]
        SM3[Slot Bank C<br/>50 Machines]
        TG1[Blackjack Pit<br/>12 Tables]
        TG2[Poker Room<br/>8 Tables]
    end

    subgraph "Edge Gateway"
        GW[Casino Floor<br/>Event Collector]
    end

    subgraph "Ingestion"
        EH[Azure Event Hub<br/>casino-slot-events<br/>32 Partitions]
    end

    subgraph "Hot Path — Real-Time"
        ADX[Azure Data Explorer<br/>Dedicated Cluster]
        KQL[KQL Continuous<br/>Queries]
        PBI[Power BI<br/>Real-Time Tiles]
        CTR[Title 31<br/>CTR Alert Engine]
    end

    subgraph "Warm Path — Batch"
        SA[Stream Analytics<br/>1-min Tumbling Window]
        ADLS[ADLS Gen2<br/>Bronze Parquet]
    end

    SM1 --> GW
    SM2 --> GW
    SM3 --> GW
    TG1 --> GW
    TG2 --> GW
    GW --> EH
    EH --> ADX
    ADX --> KQL
    KQL --> PBI
    KQL --> CTR
    EH --> SA
    SA --> ADLS

📊 Video Analytics Pipeline

graph LR
    subgraph "Cameras"
        CAM1[Floor Camera 1]
        CAM2[Floor Camera 2]
        CAM3[Floor Camera 3]
    end

    subgraph "Azure Video Analyzer"
        VA[Video Analyzer<br/>Edge Module]
        FRAME[Frame Extraction<br/>1 FPS]
    end

    subgraph "Azure Custom Vision"
        CV[Person Detection<br/>Model]
        COUNT[Occupancy<br/>Counter]
    end

    subgraph "Output"
        EH2[Event Hub<br/>density-events]
        HEAT[Heatmap<br/>Generator]
    end

    CAM1 --> VA
    CAM2 --> VA
    CAM3 --> VA
    VA --> FRAME
    FRAME --> CV
    CV --> COUNT
    COUNT --> EH2
    EH2 --> HEAT

🚀 Streaming Quick Start

# Start the slot machine event simulator
# Generates ~1,000 events/second for 150 machines
python streaming/slot_event_simulator.py \
  --event-hub-connection "$EVENTHUB_CONNECTION_STRING" \
  --machine-count 150 \
  --events-per-second 1000 \
  --denominations "0.01,0.05,0.25,1.00,5.00"

# Start the table game simulator
python streaming/table_game_simulator.py \
  --event-hub-connection "$EVENTHUB_CONNECTION_STRING" \
  --tables 20 \
  --games "blackjack,poker,baccarat,roulette"

# Deploy ADX tables and ingestion mapping
az kusto script create \
  --cluster-name casino-adx \
  --database-name gaming \
  --resource-group rg-casino-analytics \
  --script-content @streaming/adx/create_tables.kql

📊 Sample KQL — Real-Time Floor Monitoring

// Real-time floor performance by slot bank (last 15 minutes)
SlotMachineEvents
| where ingestion_time() > ago(15m)
| summarize
    total_spins = count(),
    total_wagered = sum(wager_amount),
    total_paid = sum(payout_amount),
    hold_pct = (sum(wager_amount) - sum(payout_amount)) / sum(wager_amount) * 100,
    active_machines = dcount(machine_id),
    jackpot_hits = countif(event_type == "jackpot")
    by slot_bank, bin(event_time, 1m)
| order by event_time desc, slot_bank asc

// Title 31 CTR threshold monitoring — players approaching $10K
SlotMachineEvents
| where ingestion_time() > ago(24h)
| where player_id != ""
| summarize
    total_cash_in = sum(cash_in_amount),
    total_cash_out = sum(cash_out_amount),
    session_count = dcount(session_id),
    machines_played = dcount(machine_id)
    by player_id
| where total_cash_in > 8000 or total_cash_out > 8000
| extend ctr_threshold_pct = max_of(total_cash_in, total_cash_out) / 10000.0 * 100
| order by ctr_threshold_pct desc

📎 Prerequisites

Azure Resources

  • Azure subscription with contributor access
  • Azure Data Explorer cluster (dedicated, for streaming throughput)
  • Azure Event Hub namespace (Standard or Premium, 32 partitions)
  • Azure Data Factory or Synapse Analytics
  • Azure Data Lake Storage Gen2
  • Azure SQL Database or Synapse SQL Pool
  • Azure Video Analyzer (optional, for floor density)
  • Azure Custom Vision (optional, for person detection)
  • Azure Key Vault for credentials

Tools Required

  • Azure CLI (2.55.0 or later)
  • dbt CLI (1.7.0 or later)
  • Python 3.9+
  • Git

Compliance Requirements

  • NIGC Minimum Internal Control Standards (MICS) familiarity
  • Title 31 BSA/AML compliance understanding
  • Tribal Gaming Commission authorization (for production deployment)

🚀 Quick Start

1. Environment Setup

# Clone the repository
git clone <repository-url>
cd csa-inabox/examples/casino-analytics

# Install Python dependencies
pip install -r requirements.txt

# Install dbt packages
cd domains/dbt
dbt deps

2. Generate Synthetic Data

# Generate synthetic player and transaction data
python data/generators/generate_player_data.py \
  --output-dir domains/dbt/seeds \
  --player-count 10000 \
  --days 365

python data/generators/generate_fnb_data.py \
  --output-dir domains/dbt/seeds \
  --transaction-count 500000

python data/generators/generate_hotel_data.py \
  --output-dir domains/dbt/seeds \
  --reservation-count 50000

python data/generators/generate_loyalty_data.py \
  --output-dir domains/dbt/seeds \
  --player-count 10000

3. Deploy Infrastructure

# Configure parameters
cp deploy/params.dev.json deploy/params.local.json
# Edit params.local.json with your values

# Deploy using Azure CLI
az deployment group create \
  --resource-group rg-casino-analytics \
  --template-file ../../deploy/bicep/DLZ/main.bicep \
  --parameters @deploy/params.local.json

4. Run dbt Models

cd domains/dbt

# Test connections
dbt debug

# Load seed data
dbt seed

# Run models
dbt run

# Run tests
dbt test

# Generate documentation
dbt docs generate
dbt docs serve

💡 Sample Analytics Scenarios

1. Real-Time Player Engagement Scoring

Score active players in real-time based on play velocity, bet sizing, session duration, and historical patterns to enable proactive host outreach.

-- Player engagement scores with recommended actions
SELECT
    player_id,
    player_tier,
    current_session_minutes,
    total_wagered_session,
    avg_bet_vs_historical,
    play_velocity_spins_per_min,
    win_loss_current_session,
    theoretical_win,
    engagement_score,
    churn_risk_score,
    recommended_action,
    comp_budget_available
FROM gold.gld_player_engagement
WHERE engagement_score >= 70
    AND current_session_minutes > 30
ORDER BY engagement_score DESC;

2. Floor Layout Optimization

Combine slot performance data with floor density heatmaps to optimize machine placement, bank configuration, and denomination mix.

-- Machine placement optimization recommendations
SELECT
    zone_id,
    zone_name,
    machine_count,
    avg_occupancy_pct,
    revenue_per_machine_day,
    hold_pct_actual,
    hold_pct_theoretical,
    denomination_mix_optimal,
    denomination_mix_current,
    traffic_flow_score,
    adjacency_score,
    optimization_recommendation,
    estimated_revenue_lift_pct
FROM gold.gld_floor_optimization
WHERE estimated_revenue_lift_pct > 5
ORDER BY estimated_revenue_lift_pct DESC;

3. F&B Preference Analysis

Analyze dining patterns by player tier, time of day, and gaming activity to optimize restaurant staffing, menu offerings, and comp strategies.

-- F&B preferences by player segment
SELECT
    player_tier,
    outlet_name,
    outlet_type,
    avg_check_amount,
    comp_pct,
    visits_per_month,
    preferred_day_of_week,
    preferred_time_slot,
    top_menu_categories,
    avg_time_gaming_before_dining_min,
    post_dining_return_rate_pct,
    revenue_correlation_coefficient
FROM gold.gld_fnb_preferences
ORDER BY player_tier, visits_per_month DESC;

4. Suspicious Activity Detection

Detect potential structuring, unusual play patterns, and Title 31 reporting triggers using ML-based anomaly detection.

-- Flagged suspicious activity for compliance review
SELECT
    alert_id,
    alert_date,
    player_id,
    alert_type,
    risk_score,
    total_cash_transactions_24h,
    transaction_count_24h,
    structuring_pattern_flag,
    multiple_id_flag,
    unusual_play_pattern,
    previous_alerts_count,
    recommended_action,
    ctr_required,
    sar_recommended
FROM gold.gld_suspicious_activity
WHERE alert_date >= CURRENT_DATE - INTERVAL '7 days'
    AND risk_score >= 75
ORDER BY risk_score DESC;

🔒 Regulatory Compliance

NIGC Minimum Internal Control Standards (MICS)

This platform supports MICS reporting requirements for: - Slot machine accounting (coin-in, coin-out, jackpots, fills) - Table game drop and count - Player rating accuracy verification - Cage and vault reconciliation data

Title 31 — Bank Secrecy Act / AML

Automated compliance features include: - Currency Transaction Report (CTR): Auto-detection of cash transactions ≥ $10,000 in a gaming day - Suspicious Activity Report (SAR): ML-based structuring detection (transactions just below $10K) - Multiple Transaction Log (MTL): Aggregation of cash transactions by player within 24 hours - NIGC Regulation 542: Full audit trail for all patron cash transactions

Data Retention

  • Gaming transaction data: 7 years minimum (per NIGC MICS)
  • Surveillance footage references: 7 days minimum (per MICS)
  • CTR/SAR records: 5 years (per BSA requirements)

✨ Data Products

Player Engagement (player-engagement)

  • Description: Real-time and historical player engagement scoring with churn prediction
  • Freshness: Real-time (streaming) with daily batch recalibration
  • Coverage: All rated players (synthetic: 10,000 profiles)
  • API: /api/v1/player-engagement

Floor Optimization (floor-optimization)

  • Description: Machine placement and denomination mix optimization by zone
  • Freshness: Daily recalculation with hourly density updates
  • Coverage: All gaming floor zones and machine banks
  • API: /api/v1/floor-optimization

Suspicious Activity (suspicious-activity)

  • Description: AML compliance alerts with risk scoring
  • Freshness: Real-time alerting with daily aggregation
  • Coverage: All cash transactions and player activity
  • API: /api/v1/suspicious-activity (restricted to compliance team)

⚙️ Configuration

⚙️ dbt Profiles

Add to your ~/.dbt/profiles.yml:

casino_analytics:
  target: dev
  outputs:
    dev:
      type: databricks
      host: "{{ env_var('DBT_HOST') }}"
      http_path: "{{ env_var('DBT_HTTP_PATH') }}"
      token: "{{ env_var('DBT_TOKEN') }}"
      schema: casino_dev
      catalog: dev
    prod:
      type: databricks
      host: "{{ env_var('DBT_HOST_PROD') }}"
      http_path: "{{ env_var('DBT_HTTP_PATH_PROD') }}"
      token: "{{ env_var('DBT_TOKEN_PROD') }}"
      schema: casino
      catalog: prod

⚙️ Environment Variables

# Required for streaming
EVENTHUB_CONNECTION_STRING=your-eventhub-connection

# Required for dbt
DBT_HOST=your-databricks-host
DBT_HTTP_PATH=your-sql-warehouse-path
DBT_TOKEN=your-access-token

# Optional — Video Analytics
VIDEO_ANALYZER_ENDPOINT=your-video-analyzer-endpoint
CUSTOM_VISION_ENDPOINT=your-custom-vision-endpoint
CUSTOM_VISION_KEY=your-custom-vision-key

# Optional
CASINO_LOG_LEVEL=INFO
SLOT_SIMULATOR_EVENTS_PER_SEC=1000
ADX_CLUSTER_URI=https://casino-adx.region.kusto.windows.net

🔒 Azure Government Notes

This example can be deployed to Azure Government (US) regions for tribal operations requiring government cloud compliance:

  • Use usgovvirginia or usgovarizona as your Azure region
  • ADX and Event Hub are available in Azure Government
  • Video Analyzer availability in Azure Government may vary — check regional availability
  • Custom Vision is available in Azure Government
  • Ensure NIGC regulatory data remains within approved cloud boundaries
  • Title 31 reports may need to be filed with FinCEN — verify data residency requirements

📊 Monitoring & Alerts

  • Streaming Health: Event Hub partition lag, ADX ingestion latency, event throughput
  • Title 31 Alerts: Automated notifications when players approach $10K cash threshold
  • Floor Performance: Real-time hold percentage deviation alerts by machine bank
  • Video Analytics: Camera feed health and person detection model accuracy monitoring
  • Data Quality: Automated tests on slot event integrity, player profile completeness
  • Cost Management: ADX cluster auto-scale monitoring with spend guardrails

🔧 Troubleshooting

🔧 Common Issues

  1. Event Hub Throughput: At 1,000 events/second, use at least 16 partitions. Scale to 32 for production floor loads.
  2. ADX Ingestion Lag: If ingestion falls behind, check ADX cluster compute and increase cache allocation.
  3. Slot Simulator Memory: Running 500+ virtual machines requires ~8 GB RAM. Use --batch-mode for lower memory.
  4. Video Analyzer Connectivity: Ensure edge module has outbound HTTPS access to Azure Video Analyzer endpoints.
  5. Title 31 Time Zones: Gaming day boundaries are defined by tribal gaming commission. Configure GAMING_DAY_START_HOUR appropriately (typically 6:00 AM local).
  6. Synthetic Data Realism: Adjust --house-edge and --hold-variance parameters to match your target analytics scenarios.

🔗 Contributing

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/new-data-source)
  3. Make changes and add tests
  4. Run quality checks (make lint test)
  5. Submit a pull request

🔗 License

This project is licensed under the MIT License. See LICENSE file for details.

🔗 Acknowledgments

  • National Indian Gaming Commission for regulatory framework guidance
  • Azure Cloud Scale Analytics team for the foundational platform
  • Azure Data Explorer team for high-velocity streaming capabilities
  • Contributors and the open-source community

Prerequisites / Cost / Teardown

[!IMPORTANT] Cost-safety: this vertical deploys real Azure resources. Always run teardown.sh when you are done. A forgotten workshop environment can run $200-350/day.

Prerequisites

  • Azure CLI 2.50+ logged in (az login), subscription selected (az account set --subscription <id>)
  • jq installed (used by teardown enumeration)
  • Bicep CLI 0.25+ (az bicep version)
  • Contributor + User Access Administrator on target subscription (or a pre-created RG with equivalent RBAC)
  • bash scripts/deploy/validate-prerequisites.sh passes

Cost estimate (rough, East US 2)

  • While running: ~$$200-350/day (services: Synapse, Databricks, ADX, Event Hub, IoT Hub, Cosmos DB, Storage, Key Vault)
  • Idle overnight: roughly half if you stop compute (Databricks autostop + Synapse pause)
  • Storage + Key Vault residual: <$5/month if you skip teardown

Numbers are indicative for a small demo dataset; production workloads vary significantly. Use az consumption usage list or Cost Management for live numbers.

Runtime

  • Deploy: ~45-60 minutes (first run; cold Bicep)
  • Teardown: ~15-20 minutes (async RG delete completes in the background)

Teardown

When finished, run the per-example teardown script. It enforces a typed DESTROY-casino-analytics confirmation, logs every step to reports/teardown/casino-analytics-<timestamp>.log, and deletes the resource group rg-casino-analytics along with any matching subscription-scope deployments.

# Interactive (recommended)
bash examples/casino-analytics/deploy/teardown.sh

# Dry run (enumerate only)
bash examples/casino-analytics/deploy/teardown.sh --dry-run

# From the repo root via Makefile
make teardown-example VERTICAL=casino-analytics
make teardown-example VERTICAL=casino-analytics DRYRUN=1

# CI automation (no prompt — only for ephemeral environments)
bash examples/casino-analytics/deploy/teardown.sh --yes

See docs/QUICKSTART.md#teardown for the platform-wide teardown flow.

Directory Structure

casino-analytics/
├── contracts/                # Data product contracts (schemas, SLOs, owners)
│   ├── fnb-transactions.yaml
│   ├── player-analytics.yaml
│   └── slot-events.yaml
├── data/                     # Sample data + synthetic generators
│   ├── generators/
│   └── open-data/
├── deploy/                   # Deployment parameters / Bicep templates
│   ├── params.dev.json
│   ├── params.gov.json
│   └── teardown.sh
├── domains/                  # dbt models (bronze / silver / gold) and seeds
│   └── dbt/
├── notebooks/                # Synapse / Fabric / Databricks notebooks
│   ├── floor_optimization.py
│   └── player_value_analysis.py
├── reports/                  # Power BI report templates and pbix sources
├── streaming/                # Stream Analytics jobs and KQL queries
│   ├── kql_queries.kql
│   ├── slot_event_producer.py
│   └── stream_analytics_job.json
├── video/                    # Demo recordings and storyboards
│   ├── custom_vision_setup.py
│   └── video_analyzer_config.yaml
├── ARCHITECTURE.md           # Mermaid + prose architecture diagrams
└── README.md                 # This file

Expected Results

After running the medallion pipeline against the bundled seed data, the Gold layer should populate the following tables. Row counts vary with the seed-data generator parameters; the figures below are the approximate scale you should see on a default run.

Gold Table Approximate Rows Notes
gld_compliance_monitoring TODO: capture after first run Populated from Silver via dbt --select tag:gold
gld_floor_optimization TODO: capture after first run Populated from Silver via dbt --select tag:gold
gld_player_value TODO: capture after first run Populated from Silver via dbt --select tag:gold

TODO: capture exact counts after the next end-to-end seed run. These are bounded by the seed-data generator parameters in data/generators/.