📊 Tutorial 05: Direct Lake & Power BI¶

Last Updated: 2026-04-15 | Version: 2.0 Status: ✅ Final | Maintainer: Documentation Team

🏠 Home > 📖 Tutorials > 📊 Direct Lake & Power BI

📊 Tutorial 05: Direct Lake & Power BI - Executive Analytics¶

📊 Progress Tracker¶

┌────────┬────────┬────────┬────────┬────────┬────────┬────────┬────────┬────────┬────────┐
│   00   │   01   │   02   │   03   │   04   │   05   │   06   │   07   │   08   │   09   │
│ SETUP  │ BRONZE │ SILVER │  GOLD  │   RT   │  PBI   │ PIPES  │  GOV   │ MIRROR │  AI/ML │
├────────┼────────┼────────┼────────┼────────┼────────┼────────┼────────┼────────┼────────┤
│   ✓    │   ✓    │   ✓    │   ✓    │   ✓    │  📊    │   ○    │   ○    │   ○    │   ○    │
└────────┴────────┴────────┴────────┴────────┴────────┴────────┴────────┴────────┴────────┘
                                                  ▲
                                                  │
                                             YOU ARE HERE

📋 Overview¶

This tutorial covers creating Direct Lake semantic models and Power BI reports for casino executive analytics. Direct Lake provides sub-second query performance directly against Delta tables in OneLake, combining the speed of Import mode with the freshness of DirectQuery.

💡 Why Direct Lake?

Traditional BI approaches require choosing between: - Import mode: Fast queries but stale data (requires refresh) - DirectQuery: Fresh data but slower performance

Direct Lake delivers both: Sub-second queries on always-fresh Delta table data, with no scheduled refresh required.

🎯 Learning Objectives¶

By the end of this tutorial, you will be able to:

• Understand Direct Lake mode and its benefits
• Create a semantic model from Gold layer tables
• Define table relationships for star schema
• Build DAX measures for casino KPIs
• Create executive and operational Power BI reports
• Configure row-level security (RLS)

🏗️ Direct Lake Architecture¶

Source: Direct Lake Overview in Microsoft Fabric

flowchart TB
    subgraph OneLake["☁️ OneLake"]
        GOLD[(Gold Delta Tables)]
    end

    subgraph DirectLake["🔗 Direct Lake"]
        SM[Semantic Model]
        REL[Relationships]
        DAX[DAX Measures]
    end

    subgraph PowerBI["📊 Power BI"]
        EXEC[Executive Dashboard]
        OPS[Operations Report]
        COMP[Compliance Report]
    end

    GOLD --> |Direct Lake Connection| SM
    SM --> REL
    REL --> DAX
    DAX --> EXEC
    DAX --> OPS
    DAX --> COMP

    style OneLake fill:#e3f2fd
    style DirectLake fill:#fff8e1
    style PowerBI fill:#e8f5e9

Direct Lake Benefits¶

📋 Prerequisites¶

Before starting this tutorial, ensure you have:

• ✅ Completed Tutorial 03: Gold Layer
• ✅ Gold tables populated in lh_gold Lakehouse:
• gold_slot_performance, gold_player_360, gold_compliance_reporting, gold_table_analytics, gold_financial_summary, gold_security_dashboard (fact tables from Tutorial 03)
• dim_date, dim_machine (dimensions — created by notebooks/gold/00_gold_dim_tables.py)
• gold_player_slot_daily (player-grain fact — created by notebooks/gold/07_gold_player_slot_daily.py)
• gold_player_table_daily (player-grain fact — created by notebooks/gold/08_gold_player_table_daily.py)
• ✅ Power BI license (Pro or Premium Per User)
• ✅ Fabric workspace with semantic model creation permissions

⚠️ Run the three prerequisite notebooks first. Tutorial 03 does not create dim_date, dim_machine, gold_player_slot_daily, or gold_player_table_daily. Run notebooks/gold/00_gold_dim_tables.py, notebooks/gold/07_gold_player_slot_daily.py, and notebooks/gold/08_gold_player_table_daily.py against your Silver/Gold lakehouses before proceeding. Each takes under a minute.

⚠️ License Note

Power BI Pro or Premium Per User (PPU) license is required to create and share semantic models. Users consuming reports need at least a Free license if content is in a Premium capacity.

🛠️ Step 1: Create Semantic Model¶

1.1 Create from Lakehouse¶

1. Navigate to your workspace (casino-fabric-poc)
2. Open the lh_gold Lakehouse
3. Click New semantic model in the toolbar
4. Select all 10 tables to include:

Missing tables from the picker?

• dim_date / dim_machine → run notebooks/gold/00_gold_dim_tables.py
• gold_player_slot_daily → run notebooks/gold/07_gold_player_slot_daily.py
• gold_player_table_daily → run notebooks/gold/08_gold_player_table_daily.py

1. Configure semantic model:

1. Click Create

Already created the model with fewer tables? Open sm_casino_gold → top ribbon → Home → Get data → OneLake → pick lh_gold → check any missing tables from the list above → Load. They'll join the existing tables without re-creating the model.

Direct Lake semantic models connect directly to Delta tables in OneLake for optimal performance. Source: Direct Lake overview

1.2 Verify Direct Lake Mode¶

After creation, verify the model is using Direct Lake:

1. Open the semantic model in your workspace
2. Click Settings (gear icon)
3. Verify Storage mode shows: Direct Lake
4. In the model diagram, confirm tables display "DirectLake" badge

💡 Direct Lake Requirements

For Direct Lake to work optimally: - Tables must be Delta format in OneLake - V-Order optimization should be enabled (default in Fabric) - Tables should have reasonable row counts (billions supported) - Complex calculated columns may cause DirectQuery fallback

🛠️ Step 2: Define Table Relationships¶

A well-designed star schema is critical for optimal query performance and intuitive reporting.

2.1 Open Model View¶

1. Click on the semantic model
2. Select Model view from the left panel

2.2 Create Relationships¶

erDiagram
    dim_date ||--o{ gold_slot_performance : "business_date"
    dim_date ||--o{ gold_compliance_reporting : "report_date"
    dim_date ||--o{ gold_table_analytics : "event_date"
    dim_date ||--o{ gold_financial_summary : "report_date"
    dim_date ||--o{ gold_security_dashboard : "event_date"
    dim_date ||--o{ gold_player_slot_daily : "business_date"
    dim_date ||--o{ gold_player_table_daily : "event_date"
    dim_machine ||--o{ gold_slot_performance : "machine_id"
    gold_player_360 ||--o{ gold_player_slot_daily : "player_id"
    gold_player_360 ||--o{ gold_player_table_daily : "player_id"

    dim_date {
        date date_key PK
        int year
        int quarter
        int month
        int day
        string month_name
        string day_name
        boolean is_weekend
    }

    dim_machine {
        string machine_id PK
        string manufacturer
        string machine_type
        decimal denomination
        string zone
    }

    gold_player_360 {
        string player_id PK
        string loyalty_tier
        boolean vip_flag
        string churn_risk
        decimal player_value_score
    }

    gold_slot_performance {
        date business_date FK
        string machine_id FK
        int unique_players
        decimal total_coin_in
        decimal total_coin_out
        decimal hold_percentage
    }

    gold_table_analytics {
        date event_date FK
        string game_type
        int unique_players
        decimal total_drop
        decimal table_win
    }

    gold_player_slot_daily {
        string player_id FK
        date business_date FK
        decimal total_coin_in
        decimal net_win
        int total_games_played
        int jackpot_count
    }

    gold_player_table_daily {
        string player_id FK
        date event_date FK
        decimal total_drop
        decimal player_net_win
        int total_hands_played
        decimal total_hours_played
    }

    gold_compliance_reporting {
        date report_date FK
        int ctr_count
        int sar_count
        int w2g_count
    }

    gold_financial_summary {
        date report_date FK
        decimal gross_revenue
        decimal total_drop
    }

    gold_security_dashboard {
        date event_date FK
        string zone
        int incident_count
    }

2.3 Configure Each Relationship¶

Create the 10 relationships below by dragging columns between tables (or via Manage relationships → New). The Many side is always the Fact; the One side is always the Dimension.

Dimension → Fact relationships (dim_date as universal date dimension)¶

Machine dimension → Slot fact¶

Player dimension → Player-grain facts¶

gold_player_360 now joins the model through the two player-grain fact tables (gold_player_slot_daily, gold_player_table_daily). This enables cross-filtering VIP/loyalty/churn slicers against slot and table activity.

Configuration for every relationship¶

• Cardinality: Many-to-One (from fact to dimension)
• Cross-filter direction: Single
• Make this relationship active: Yes

Ambiguity note. Because dim_date filters five facts and several of those facts share natural-key columns (e.g., report_date appears on both gold_compliance_reporting and gold_financial_summary), Power BI may warn about "ambiguous paths." This is expected and safe — each relationship is between distinct tables. If you see a warning about an inactive relationship, that's fine; for a Direct Lake star schema, one active date relationship per fact is what you want.

⚠️ Relationship Best Practices. Always use single cross-filter direction for better performance. Avoid bidirectional filters unless absolutely necessary. Use integer keys when possible for faster joins. Validate relationships with sample queries.

🛠️ Step 3: Create DAX Measures¶

DAX measures provide calculated metrics that respond to filter context in reports.

3.1 Create Measures Table¶

1. In the model view, click New measure
2. Create a "Measures" display folder to organize measures

Measures appear in the Data pane with a calculator icon. Source: Create measures in Power BI Desktop

The formula bar displays the DAX expression for the selected measure. Source: Create measures in Power BI Desktop

3.2 Slot Performance Measures¶

// ===========================================
// SLOT PERFORMANCE MEASURES
// ===========================================

// Total Coin In
Total Coin In =
SUM(gold_slot_performance[total_coin_in])

// Total Coin Out
Total Coin Out =
SUM(gold_slot_performance[total_coin_out])

// Net Win (House Win)
Net Win =
[Total Coin In] - [Total Coin Out]

// Hold Percentage
Hold % =
DIVIDE(
    [Net Win],
    [Total Coin In],
    0
) * 100

// Theoretical Win (based on programmed hold)
Theoretical Win =
SUMX(
    gold_slot_performance,
    gold_slot_performance[total_coin_in] *
    gold_slot_performance[avg_theoretical_hold]
)

// Hold Variance (Actual vs Theoretical)
Hold Variance =
[Net Win] - [Theoretical Win]

// Hold Variance % (for alerting)
Hold Variance % =
DIVIDE(
    [Hold Variance],
    [Theoretical Win],
    0
) * 100

// Total Games Played
Total Games =
SUM(gold_slot_performance[total_games])

// Unique Players
Unique Players =
SUM(gold_slot_performance[unique_players])

// Average Bet
Avg Bet =
DIVIDE(
    [Total Coin In],
    [Total Games],
    0
)

// Win Per Machine
Win Per Machine =
DIVIDE(
    [Net Win],
    DISTINCTCOUNT(gold_slot_performance[machine_id]),
    0
)

// Active Machines Count
Active Machines =
DISTINCTCOUNT(gold_slot_performance[machine_id])

3.3 Player Analytics Measures¶

// ===========================================
// PLAYER MEASURES
// ===========================================

// Total Players
Total Players =
COUNTROWS(gold_player_360)

// VIP Player Count
VIP Players =
CALCULATE(
    COUNTROWS(gold_player_360),
    gold_player_360[vip_flag] = TRUE()
)

// VIP Percentage
VIP % =
DIVIDE(
    [VIP Players],
    [Total Players],
    0
) * 100

// High Churn Risk Players
High Churn Risk Players =
CALCULATE(
    COUNTROWS(gold_player_360),
    gold_player_360[churn_risk] = "High"
)

// Average Player Value Score
Avg Player Value =
AVERAGE(gold_player_360[player_value_score])

// Total Player Theo
Total Player Theo =
SUM(gold_player_360[total_theo_win])

// Active Players (visited in last 30 days)
Active Players 30D =
CALCULATE(
    COUNTROWS(gold_player_360),
    gold_player_360[days_since_visit] <= 30
)

// Lapsed Players (no visit in 90+ days)
Lapsed Players =
CALCULATE(
    COUNTROWS(gold_player_360),
    gold_player_360[days_since_visit] > 90
)

// Player Retention Rate
Retention Rate 30D =
DIVIDE(
    [Active Players 30D],
    [Total Players],
    0
) * 100

3.4 Compliance Measures¶

// ===========================================
// COMPLIANCE MEASURES
// ===========================================

// Total CTR Filings
CTR Count =
SUM(gold_compliance_reporting[ctr_count])

// Total SAR Filings
SAR Count =
SUM(gold_compliance_reporting[sar_count])

// Total W2G Filings
W2G Count =
SUM(gold_compliance_reporting[w2g_count])

// Total Regulatory Filings
Total Filings =
[CTR Count] + [SAR Count] + [W2G Count]

// CTR Total Amount
CTR Total Amount =
SUM(gold_compliance_reporting[ctr_total_amount])

// Average Daily Filing Rate
Daily Filing Rate =
DIVIDE(
    [Total Filings],
    DISTINCTCOUNT(gold_compliance_reporting[report_date]),
    0
)

// SAR Rate (per 1000 players)
SAR Rate =
DIVIDE(
    [SAR Count],
    [Total Players] / 1000,
    0
)

3.5 Time Intelligence Measures¶

// ===========================================
// TIME INTELLIGENCE MEASURES
// ===========================================

// Coin In - Month to Date
Coin In MTD =
TOTALMTD(
    [Total Coin In],
    dim_date[date_key]
)

// Coin In - Year to Date
Coin In YTD =
TOTALYTD(
    [Total Coin In],
    dim_date[date_key]
)

// Net Win - Previous Month
Net Win PM =
CALCULATE(
    [Net Win],
    DATEADD(dim_date[date_key], -1, MONTH)
)

// Net Win - Month over Month Growth %
Net Win MoM % =
VAR CurrentMonth = [Net Win]
VAR PreviousMonth = [Net Win PM]
RETURN
DIVIDE(
    CurrentMonth - PreviousMonth,
    PreviousMonth,
    0
) * 100

// Net Win - Previous Year
Net Win PY =
CALCULATE(
    [Net Win],
    SAMEPERIODLASTYEAR(dim_date[date_key])
)

// Net Win - Year over Year Growth %
Net Win YoY % =
VAR CurrentPeriod = [Net Win]
VAR PreviousYear = [Net Win PY]
RETURN
DIVIDE(
    CurrentPeriod - PreviousYear,
    PreviousYear,
    0
) * 100

// 7-Day Rolling Average Coin In
Coin In 7D Avg =
AVERAGEX(
    DATESINPERIOD(
        dim_date[date_key],
        MAX(dim_date[date_key]),
        -7,
        DAY
    ),
    [Total Coin In]
)

// 30-Day Rolling Average Net Win
Net Win 30D Avg =
AVERAGEX(
    DATESINPERIOD(
        dim_date[date_key],
        MAX(dim_date[date_key]),
        -30,
        DAY
    ),
    [Net Win]
)

💡 DAX Best Practices

• Use measures over calculated columns for better performance
• Avoid FILTER() with large tables - use simpler predicates
• Use variables (VAR) to improve readability and avoid recalculation
• Test measures with different filter contexts before deploying

🛠️ Step 4: Create Executive Dashboard¶

4.1 Create New Report¶

1. From the semantic model, click Create report
2. Or in Power BI service: + New > Report > Select Casino Analytics Model
3. The report canvas opens in edit mode

4.2 Page 1: Executive Summary¶

Layout Design¶

┌─────────────────────────────────────────────────────────────────────────────────┐
│  🎰 CASINO EXECUTIVE DASHBOARD                        📅 Date: [Date Slicer]   │
├──────────────┬──────────────┬──────────────┬──────────────┬────────────────────┤
│   💰         │    📊        │    👥        │    🎮        │   🏷️ Zone Filter   │
│   $2.5M      │    8.2%      │   12,450     │    1.2M      │   [Dropdown]       │
│   Net Win    │   Hold %     │   Players    │   Games      │                    │
│   ▲ +5.2%    │   ▼ -0.3%    │   ▲ +2.1%    │   ▲ +8.4%    │                    │
├──────────────┴──────────────┴──────────────┴──────────────┴────────────────────┤
│                         📈 NET WIN TREND (30 Days)                              │
│  ████████████████████████████████████████████████████████████████████████████  │
│  [Line Chart - Daily Net Win with trend line and target]                        │
├─────────────────────────────────────────┬───────────────────────────────────────┤
│   📊 PERFORMANCE BY ZONE                │   🏆 TOP 10 MACHINES                  │
│   ┌─────────────────────────────┐       │   ┌─────────────────────────────────┐ │
│   │ ████████████ Main Floor     │       │   │ Machine  │ Zone   │ Net Win    │ │
│   │ ██████████ High Limit       │       │   │ SM-0234  │ VIP    │ $45,230    │ │
│   │ ████████ VIP                │       │   │ SM-0089  │ HL     │ $38,420    │ │
│   │ ██████ Penny Palace         │       │   │ SM-0445  │ Main   │ $32,100    │ │
│   └─────────────────────────────┘       │   └─────────────────────────────────┘ │
├─────────────────────────────────────────┼───────────────────────────────────────┤
│   📋 COMPLIANCE SUMMARY                 │   👥 PLAYER TIER DISTRIBUTION         │
│   ┌────────┬────────┬────────┐          │        ┌─────────────┐                │
│   │  CTR   │  SAR   │  W2G   │          │       ╱  Platinum   ╲                │
│   │   45   │   12   │  234   │          │      ╱    Gold       ╲               │
│   │ $4.2M  │        │ $1.8M  │          │     ╱     Silver      ╲              │
│   └────────┴────────┴────────┘          │    ╱      Bronze       ╲             │
│                                          │        [Donut Chart]                 │
└─────────────────────────────────────────┴───────────────────────────────────────┘

Create Visuals¶

Drag fields from the Data pane to create visualizations on the report canvas. Source: Report view in Power BI Desktop

Visualizations powered by DAX measures for business analytics. Source: Create measures in Power BI Desktop

1. KPI Cards (Top Row)

Create 4 card visuals with conditional formatting:

2. Net Win Trend (Line Chart)

3. Performance by Zone (Bar Chart)

4. Top 10 Machines (Table)

Apply Top N filter: Top 10 by Net Win

5. Compliance Summary (Multi-row Card)

6. Player Tier Distribution (Donut Chart)

4.3 Page 2: Slot Operations¶

Layout Design¶

┌─────────────────────────────────────────────────────────────────────────────────┐
│  🎰 SLOT OPERATIONS                     📅 Date: [Range]  🏷️ Zone: [Multi]     │
├─────────────────────────────────────────────────────────────────────────────────┤
│                    📊 MACHINE PERFORMANCE MATRIX                                 │
│  ┌───────────────────────────────────────────────────────────────────────────┐  │
│  │ Zone/Denom  │  $0.01  │  $0.05  │  $0.25  │  $1.00  │  $5.00  │  Total   │  │
│  ├─────────────┼─────────┼─────────┼─────────┼─────────┼─────────┼──────────┤  │
│  │ Main Floor  │  $125K  │  $89K   │  $234K  │  $456K  │  $78K   │  $982K   │  │
│  │ High Limit  │    -    │    -    │  $45K   │  $234K  │  $567K  │  $846K   │  │
│  └───────────────────────────────────────────────────────────────────────────┘  │
├─────────────────────────────────────────┬───────────────────────────────────────┤
│   📈 HOURLY ACTIVITY PATTERN            │   📊 HOLD VARIANCE ANALYSIS           │
│   ┌─────────────────────────────┐       │   ┌─────────────────────────────────┐ │
│   │ [Area chart by hour]        │       │   │ [Scatter: Theo vs Actual]       │ │
│   │                             │       │   │   x: Theoretical Win             │ │
│   │ Peak: 8PM-11PM              │       │   │   y: Actual Win                  │ │
│   │ Low: 4AM-7AM                │       │   │   Color by variance              │ │
│   └─────────────────────────────┘       │   └─────────────────────────────────┘ │
├─────────────────────────────────────────┼───────────────────────────────────────┤
│   🏭 MANUFACTURER PERFORMANCE           │   🎰 JACKPOT SUMMARY                  │
│   ┌─────────────────────────────┐       │   ┌─────────────────────────────────┐ │
│   │ [Clustered bar chart]       │       │   │ Total Jackpots: 234             │ │
│   │   - IGT                     │       │   │ Total Amount: $1.2M             │ │
│   │   - Aristocrat              │       │   │ Largest: $45,230 (SM-0234)      │ │
│   │   - Scientific Games        │       │   │ [Table of recent jackpots]      │ │
│   └─────────────────────────────┘       │   └─────────────────────────────────┘ │
└─────────────────────────────────────────┴───────────────────────────────────────┘

Key Visuals¶

1. Performance Matrix - Visual type: Matrix - Rows: Zone - Columns: Denomination - Values: Net Win, Hold %, Games

2. Hold Variance Scatter Plot - X-axis: Theoretical Win - Y-axis: Net Win (Actual) - Size: Total Games - Color saturation: Hold Variance % - Reference line: y=x (where actual = theoretical)

4.4 Page 3: Player Analytics¶

Layout Design¶

┌─────────────────────────────────────────────────────────────────────────────────┐
│  👥 PLAYER ANALYTICS                           🏷️ Tier: [Slicer] 📅 [Date]     │
├──────────────┬──────────────┬──────────────┬────────────────────────────────────┤
│   👥         │   ⭐         │    ⚠️        │    📈                               │
│   45,230     │   2,340      │    1,256     │    72.5                            │
│   Total      │   VIP        │   At Risk    │    Avg Value                       │
│   Players    │   Players    │   (Churn)    │    Score                           │
├──────────────┴──────────────┴──────────────┴────────────────────────────────────┤
│                    📊 PLAYER VALUE DISTRIBUTION                                  │
│  ┌───────────────────────────────────────────────────────────────────────────┐  │
│  │ [Histogram of player_value_score with normal curve overlay]               │  │
│  └───────────────────────────────────────────────────────────────────────────┘  │
├─────────────────────────────────────────┬───────────────────────────────────────┤
│   📊 TIER BREAKDOWN                     │   ⚠️ CHURN RISK ANALYSIS             │
│   ┌─────────────────────────────┐       │   ┌─────────────────────────────────┐ │
│   │ [Stacked bar by tier]       │       │   │ [Pie: High/Medium/Low risk]     │ │
│   │   Platinum: 5%              │       │   │                                 │ │
│   │   Gold: 15%                 │       │   │   High: 12%                     │ │
│   │   Silver: 30%               │       │   │   Medium: 28%                   │ │
│   │   Bronze: 50%               │       │   │   Low: 60%                      │ │
│   └─────────────────────────────┘       │   └─────────────────────────────────┘ │
├─────────────────────────────────────────────────────────────────────────────────┤
│                    🏆 TOP PLAYERS BY THEORETICAL VALUE                          │
│  ┌───────────────────────────────────────────────────────────────────────────┐  │
│  │ Player ID  │  Tier     │  Total Theo  │  Visits  │  Churn Risk  │ Action  │  │
│  │ PLY-00234  │ Platinum  │   $125,000   │   45     │    Low       │  [View] │  │
│  │ PLY-00892  │ Platinum  │   $98,500    │   38     │    Medium    │  [View] │  │
│  └───────────────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────────┘

🛠️ Step 5: Configure Report Settings¶

5.1 Publish Report¶

1. Save the report: File > Save as > Casino Executive Dashboard
2. Click Publish
3. Select workspace: casino-fabric-poc

5.2 Verify Direct Lake Refresh¶

For Direct Lake semantic models, data updates are automatic:

1. Open the report in Power BI service
2. Click Settings (gear icon) on the semantic model
3. Verify under Refresh:
4. Shows: "Direct Lake - data is always current"
5. No scheduled refresh is needed

💡 Direct Lake Refresh Behavior

• Data changes in Delta tables are visible within seconds
• No manual refresh button needed
• If framing (metadata sync) is needed, it happens automatically
• Monitor with semantic model refresh history for any issues

5.3 Configure Row-Level Security (RLS)¶

Restrict data access by user role (e.g., zone managers see only their zones).

Create RLS Role¶

1. Open semantic model in Edit mode
2. Go to Modeling > Manage roles
3. Create new role: Zone Manager

Define DAX Filter¶

// RLS Rule: Zone-based access
// Users see only their assigned zone(s)
[Zone Security] =
VAR UserZones =
    LOOKUPVALUE(
        UserZoneMapping[Zone],
        UserZoneMapping[UserEmail],
        USERPRINCIPALNAME()
    )
RETURN
    gold_slot_performance[zone] = UserZones
    || UserZones = "All"  // "All" grants full access

Alternative: Multiple Zones per User¶

// RLS Rule: Support multiple zones per user
[Zone Security Multi] =
VAR UserEmail = USERPRINCIPALNAME()
VAR AllowedZones =
    CALCULATETABLE(
        VALUES(UserZoneMapping[Zone]),
        UserZoneMapping[UserEmail] = UserEmail
    )
RETURN
    gold_slot_performance[zone] IN AllowedZones
    || "All" IN AllowedZones

Test RLS¶

1. Click View as role in Power BI Desktop
2. Select Zone Manager role
3. Enter test user email
4. Verify data is filtered correctly

🛠️ Step 6: Create Paginated Report (Optional)¶

For compliance reporting requiring exact formatting and pagination.

6.1 Use Power BI Report Builder¶

1. Download Power BI Report Builder (free)
2. Create new report connecting to semantic model
3. Design compliance-friendly layouts:
4. CTR summary with transaction details
5. SAR filing log with timestamps
6. W2G jackpot listing with IRS form format

6.2 Paginated Report Features¶

✅ Validation Checklist¶

Before moving to the next tutorial, verify:

• Direct Lake Semantic Model Created - Model exists with Direct Lake storage mode
• Reports Load Successfully - Executive dashboard opens without errors
• Visuals Display Data - All charts and tables show current data
• Relationships Work - Drill-through and filters function correctly
• DAX Measures Calculate - All KPIs return expected values
• Auto-Refresh Works - Data changes appear within seconds

1. Navigate to workspace
2. Look for semantic model (dataset icon)
3. Click on model > Settings
4. Under "Storage mode" section, verify:
   ✅ "Direct Lake" is selected
   ✅ Source lakehouse shows "lh_gold"
   ✅ Status shows "Connected"

1. Open "Casino Executive Dashboard" report
2. Verify:
   - Report loads without error messages
   - All pages accessible (Executive Summary, Slot Performance, Player Insights)
   - No "Can't load visual" errors
   - Data loads within 5-10 seconds

Check each visual on each page:

Page 1 - Executive Summary:
✅ Total Revenue card shows value
✅ Hold % gauge displays percentage
✅ Daily trend line chart shows data points
✅ Top machines table populated

Page 2 - Slot Performance:
✅ Machine heatmap shows color gradients
✅ Performance matrix has data
✅ Trend over time chart displays lines

Page 3 - Player Insights:
✅ Player segmentation chart shows segments
✅ Value tier distribution populated
✅ Top players table shows names/metrics

Test relationship functionality:

1. Click-through test:
   - Click a zone in bar chart
   - Verify all other visuals filter to that zone

2. Slicer test:
   - Select date range in date slicer
   - Verify all visuals update for that period

3. Drill-through test:
   - Right-click a machine in the table
   - Select "Drill through" > Machine Details
   - Verify detail page shows correct machine data

Verify key DAX measures in report:

✅ [Total Coin In] - Should be > $0
✅ [Total Net Win] - Should be positive
✅ [Hold %] - Should be between 5-15%
✅ [Theoretical Win] - Should be > $0
✅ [Variance to Theo] - Should be ± reasonable
✅ [Player Count] - Should be > 0

Test a measure manually:
- Select a measure in Fields pane
- Click "Table view" in Modeling tab
- Verify it calculates without errors

# 1. Update Gold layer data
from pyspark.sql.functions import col, lit

# Add a test record with recognizable values
df = spark.table("lh_gold.gold_slot_performance").limit(1)
test_record = df.withColumn("net_win", lit(99999.99))
test_record.write.format("delta").mode("append").saveAsTable("lh_gold.gold_slot_performance")

print("✅ Test record added with net_win = 99999.99")

2. In Power BI:
   - Open the report (or refresh if already open)
   - Navigate to Slot Performance page
   - Look for the test machine
   - Verify net_win shows 99999.99
   - Data should appear within 5-10 seconds (no manual refresh needed!)

Test query performance:

1. Open report in Power BI Desktop or Service
2. Enable Performance Analyzer (View > Performance Analyzer)
3. Click "Start recording"
4. Refresh all visuals (Ctrl+R)
5. Click "Stop recording"

Expected performance:
✅ Visual queries: < 2 seconds each
✅ DAX queries: < 1 second for most measures
✅ Page load: < 5 seconds

🔧 Performance Optimization¶

For Direct Lake¶

For Reports¶

Monitoring Direct Lake¶

Check the semantic model for fallback indicators:

// Query to identify fallback scenarios
// Run in DAX Studio or external tools
EVALUATE
INFO.STORAGETABLECOLUMNS()

🔧 Troubleshooting¶

Issue: Fallback to DirectQuery Mode¶

Common fallback triggers: - Complex calculated columns with RELATED() - Very large tables (check row counts) - Certain DAX functions on large datasets

Issue: Slow Query Performance¶

Issue: Data Not Refreshing¶

🎉 Summary¶

Congratulations! You have successfully built executive analytics using Direct Lake and Power BI.

What You Accomplished¶

• ✅ Created a Direct Lake semantic model for casino analytics
• ✅ Defined table relationships for star schema
• ✅ Built DAX measures for slot, player, and compliance KPIs
• ✅ Created time intelligence calculations (MTD, YTD, trends)
• ✅ Designed executive dashboards with automatic data refresh
• ✅ Configured row-level security for zone-based access

Key Takeaways¶

Architecture Achieved¶

flowchart LR
    subgraph Data["🗄️ Data Layer"]
        DT[Delta Tables<br/>Gold Layer]
    end

    subgraph Semantic["📊 Semantic Layer"]
        DL[Direct Lake<br/>Semantic Model]
    end

    subgraph Reporting["📈 Reporting Layer"]
        EXEC[Executive<br/>Dashboard]
        OPS[Operations<br/>Report]
        COMP[Compliance<br/>Report]
    end

    DT --> |Auto-sync| DL
    DL --> EXEC
    DL --> OPS
    DL --> COMP

    style Data fill:#e3f2fd
    style Semantic fill:#fff8e1
    style Reporting fill:#e8f5e9

🚀 Next Steps¶

Continue your learning journey:

Next Tutorial: Tutorial 06: Data Pipelines - Orchestrate end-to-end data workflows

Optional Deep Dives: - Add more sophisticated DAX calculations - Create mobile-optimized report layouts - Implement Q&A natural language queries - Set up email subscriptions and alerts

📚 Resources¶

🧭 Navigation¶

⬆️ Back to Top | 📚 Tutorials | 🏠 Home