Agent Skills: Tableau Skill

Master Tableau development including calculated fields, LOD expressions, table calculations, and dashboard design

UncategorizedID: pluginagentmarketplace/custom-plugin-bi-analyst/tableau

Skill Files

Browse the full folder contents for tableau.

Download Skill

Loading file tree…

skills/tableau/SKILL.md

Skill Metadata

Name
tableau
Description
Master Tableau development including calculated fields, LOD expressions, table calculations, and dashboard design

Tableau Skill

Master Tableau development including calculated fields, LOD expressions, table calculations, dashboard actions, and performance optimization.

Quick Start (5 minutes)

// 3 essential Tableau patterns:

// 1. Basic calculated field
[Profit Margin] = [Profit] / [Sales]

// 2. LOD Expression (FIXED)
[Customer First Order] = { FIXED [Customer ID] : MIN([Order Date]) }

// 3. Table Calculation
[Running Total] = RUNNING_SUM(SUM([Sales]))

Core Concepts

Order of Operations

┌─────────────────────────────────────────────────────────────┐
│                 TABLEAU ORDER OF OPERATIONS                  │
├─────────────────────────────────────────────────────────────┤
│  1. Extract Filters (data source)                           │
│  2. Data Source Filters                                     │
│  3. Context Filters          ◄── Set BEFORE LOD             │
│  4. FIXED LOD Expressions    ◄── Computed here              │
│  5. Set/Top N Filters                                       │
│  6. Dimension Filters                                       │
│  7. INCLUDE/EXCLUDE LOD      ◄── Computed here              │
│  8. Measure Filters                                         │
│  9. Table Calculations       ◄── Computed last              │
│ 10. Trend Lines, Reference Lines                            │
└─────────────────────────────────────────────────────────────┘

LOD Expression Types

┌──────────────────────────────────────────────────────────────┐
│                    LOD EXPRESSIONS                           │
├──────────────────────────────────────────────────────────────┤
│                                                              │
│  FIXED: Calculate at SPECIFIED level                         │
│  { FIXED [Customer] : SUM([Sales]) }                        │
│  → Always at customer level, regardless of view              │
│                                                              │
│  INCLUDE: Add dimension to calculation                       │
│  { INCLUDE [Product] : AVG([Price]) }                       │
│  → Include product even if not in view                      │
│                                                              │
│  EXCLUDE: Remove dimension from calculation                  │
│  { EXCLUDE [Region] : SUM([Sales]) }                        │
│  → Calculate without region grouping                        │
│                                                              │
└──────────────────────────────────────────────────────────────┘

Table Calculation Addressing

Compute Using: Controls HOW the calculation traverses

Table (across)     → Left to right
Table (down)       → Top to bottom
Table (across then down) → Left-right, then next row
Pane (across)      → Within each pane left-right
Cell               → Each cell individually
Specific Dimensions → Custom control

PARTITIONING: Dimensions that DEFINE the scope
ADDRESSING: Dimensions that TRAVERSE for calculation

Code Examples

LOD Expressions

// Customer's First Purchase Date
[Customer First Order] =
{ FIXED [Customer ID] : MIN([Order Date]) }

// Customer Lifetime Value
[Customer LTV] =
{ FIXED [Customer ID] : SUM([Sales]) }

// Is New Customer (Boolean)
[Is New Customer] =
[Order Date] = [Customer First Order]

// Days Since First Order
[Customer Tenure] =
DATEDIFF('day', [Customer First Order], [Order Date])

// Percent of Customer Total
[% of Customer Sales] =
SUM([Sales]) / [Customer LTV]

// Cohort Month
[Cohort Month] =
{ FIXED [Customer ID] : MIN(DATETRUNC('month', [Order Date])) }

// Average Order Value per Customer (then aggregate)
[Avg Order Value per Customer] =
{ FIXED [Customer ID] : AVG([Sales]) }
// Use AVG([Avg Order Value per Customer]) to get average across customers

// New vs Returning Segment
[Customer Type] =
IF [Order Date] = [Customer First Order]
THEN "New"
ELSE "Returning"
END

Table Calculations

// Running Total
[Running Total Sales] =
RUNNING_SUM(SUM([Sales]))
// Compute using: specific date dimension

// Year-over-Year Growth
[YoY Growth] =
(SUM([Sales]) - LOOKUP(SUM([Sales]), -1)) / ABS(LOOKUP(SUM([Sales]), -1))
// Compute using: Year

// Percent of Total
[% of Total] =
SUM([Sales]) / TOTAL(SUM([Sales]))

// Rank
[Sales Rank] =
RANK(SUM([Sales]))
// Compute using: Product

// Moving Average (3 periods)
[3-Period MA] =
WINDOW_AVG(SUM([Sales]), -2, 0)

// Cumulative Percent (Pareto)
[Cumulative %] =
RUNNING_SUM(SUM([Sales])) / TOTAL(SUM([Sales]))

// Index (normalized 0-100)
[Sales Index] =
(SUM([Sales]) - WINDOW_MIN(SUM([Sales]))) /
(WINDOW_MAX(SUM([Sales])) - WINDOW_MIN(SUM([Sales]))) * 100

// Period-over-Period Comparison
[MoM Change] =
ZN(SUM([Sales])) - LOOKUP(ZN(SUM([Sales])), -1)

Advanced Calculated Fields

// Fiscal Year (April Start)
[Fiscal Year] =
IF MONTH([Order Date]) >= 4
THEN YEAR([Order Date])
ELSE YEAR([Order Date]) - 1
END

// Business Days Between Dates (excluding weekends)
[Business Days] =
DATEDIFF('day', [Start Date], [End Date])
- (DATEDIFF('week', [Start Date], [End Date]) * 2)
- IF DATEPART('weekday', [Start Date]) = 1 THEN 1 ELSE 0 END
- IF DATEPART('weekday', [End Date]) = 7 THEN 1 ELSE 0 END

// Dynamic Date Granularity
[Dynamic Date] =
CASE [Date Granularity Parameter]
    WHEN 'Day' THEN STR(DAY([Order Date])) + ' ' + DATENAME('month', [Order Date])
    WHEN 'Week' THEN 'Week ' + STR(DATEPART('week', [Order Date]))
    WHEN 'Month' THEN DATENAME('month', [Order Date]) + ' ' + STR(YEAR([Order Date]))
    WHEN 'Quarter' THEN 'Q' + STR(DATEPART('quarter', [Order Date])) + ' ' + STR(YEAR([Order Date]))
    WHEN 'Year' THEN STR(YEAR([Order Date]))
END

// Safe Division with NULL handling
[Safe Margin] =
IF ISNULL([Revenue]) OR [Revenue] = 0
THEN NULL
ELSE ([Revenue] - [Cost]) / [Revenue]
END

Best Practices

Calculation Naming Convention

Prefix by type:
• agg_    : Aggregate calculations
• lod_    : LOD expressions
• tc_     : Table calculations
• param_  : Parameter-dependent
• bool_   : Boolean/filter fields
• calc_   : General calculations

Examples:
• agg_Total Sales
• lod_Customer First Order
• tc_Rank by Category
• param_Selected Date Range
• bool_Is Current Year
• calc_Profit Margin

Performance Optimization

✓ DO:
• Use extracts for large data
• Use context filters for LOD
• Aggregate at data source level
• Limit marks to <10,000 per view
• Use data source filters

✗ DON'T:
• Nest multiple LODs deeply
• Use live connections for large data
• Create complex string calculations
• Use CONTAINS for exact matches
• Include unused fields in view

Filter Best Practices

1. Context Filters: Use when LOD needs filtering
2. Data Source Filters: Apply at connection level
3. Extract Filters: Reduce extract size
4. Dimension Filters: Most common, after LOD
5. Measure Filters: After aggregation
6. Table Calc Filters: Applied last

Common Patterns

Dashboard Actions Configuration

Filter Action:
  name: "Select Category"
  source_sheet: "Category Overview"
  target_sheets:
    - "Product Detail"
    - "Trend Chart"
  run_on: "Select"
  clearing: "Show all values"
  source_fields: ["Category"]
  target_fields: ["Category"]

Highlight Action:
  name: "Highlight Region"
  source_sheet: "Regional Map"
  target_sheets: "All sheets in dashboard"
  run_on: "Hover"
  target_highlighting: true

URL Action:
  name: "Open Product Page"
  source_sheet: "Product Table"
  run_on: "Menu"
  url: "https://products.example.com/<Product ID>"
  url_target: "New Tab"

Parameter Action:
  name: "Set Selected Customer"
  source_sheet: "Customer List"
  run_on: "Select"
  source_field: "Customer ID"
  target_parameter: "Selected Customer"

Cohort Analysis Setup

// Step 1: Create cohort dimension
[Cohort Month] = { FIXED [Customer ID] : MIN(DATETRUNC('month', [Order Date])) }

// Step 2: Calculate period number
[Periods Since Cohort] = DATEDIFF('month', [Cohort Month], DATETRUNC('month', [Order Date]))

// Step 3: Create retention calculation (as table calc)
[Retention Rate] = COUNTD([Customer ID]) / LOOKUP(COUNTD([Customer ID]), FIRST())
// Compute using: Periods Since Cohort

// Layout:
// Rows: Cohort Month
// Columns: Periods Since Cohort
// Color: Retention Rate

Dynamic Parameter Selector

// Parameter: Select Measure
// Allowable values: Revenue, Profit, Orders

// Calculated field using parameter
[Selected Measure] =
CASE [Select Measure Parameter]
    WHEN 'Revenue' THEN SUM([Sales])
    WHEN 'Profit' THEN SUM([Profit])
    WHEN 'Orders' THEN COUNTD([Order ID])
END

// Dynamic title
[Chart Title] = "Performance by Region: " + [Select Measure Parameter]

Retry Logic

const publishWorkbook = async (workbook: Workbook) => {
  const retryConfig = {
    maxRetries: 3,
    backoffMs: [5000, 15000, 30000]
  };

  for (let attempt = 0; attempt <= retryConfig.maxRetries; attempt++) {
    try {
      return await tableauServer.publish(workbook);
    } catch (error) {
      if (attempt === retryConfig.maxRetries) throw error;
      if (error.code === 'CONFLICT') {
        await resolveConflict(workbook);
      }
      await sleep(retryConfig.backoffMs[attempt]);
    }
  }
};

Logging Hooks

const tableauHooks = {
  onCalcCreate: (calcName, type) => {
    console.log(`[TABLEAU] Created ${type}: ${calcName}`);
  },

  onQueryExecute: (datasource, rowCount, duration) => {
    console.log(`[TABLEAU] Query on ${datasource}: ${rowCount} rows in ${duration}ms`);
    if (duration > 5000) {
      console.warn(`[TABLEAU] Slow query on ${datasource}`);
    }
  },

  onExtractRefresh: (extractName, status) => {
    console.log(`[TABLEAU] Extract ${extractName}: ${status}`);
  }
};

Unit Test Template

describe('Tableau Skill', () => {
  describe('LOD Expressions', () => {
    it('should calculate customer first order correctly', () => {
      const result = evaluateLOD('{ FIXED [Customer ID] : MIN([Order Date]) }');
      expect(result['CUST001']).toBe('2024-01-15');
    });
  });

  describe('Table Calculations', () => {
    it('should compute running total', () => {
      const data = [100, 200, 150];
      const result = runningSum(data);
      expect(result).toEqual([100, 300, 450]);
    });

    it('should rank correctly with ties', () => {
      const data = [100, 200, 200, 150];
      const result = rank(data, 'desc');
      expect(result).toEqual([4, 1, 1, 3]);
    });
  });

  describe('Filters', () => {
    it('should respect context filter order', () => {
      const result = evaluateWithFilters({
        context: [{ field: 'Region', value: 'West' }],
        lod: '{ FIXED [State] : SUM([Sales]) }'
      });
      expect(result).toContainStates(['CA', 'WA', 'OR']);
    });
  });
});

Troubleshooting

| Issue | Cause | Solution | |-------|-------|----------| | Cannot mix aggregate | Aggregate/row mismatch | Wrap with ATTR() or aggregate | | LOD returns wrong values | Filter order | Use context filter | | Table calc addressing wrong | Compute using incorrect | Manually set addressing | | Slow dashboard | Too many marks | Reduce detail, use extract | | Extract fails | Data type mismatch | Check null handling |

Resources

  • Tableau Help: Official calculation reference
  • LOD Expressions Deep Dive: Tableau whitepaper
  • Performance Best Practices: Tableau documentation
  • The Information Lab: Advanced tutorials

Version History

| Version | Date | Changes | |---------|------|---------| | 1.0.0 | 2024-01 | Initial release | | 2.0.0 | 2025-01 | Production-grade with LOD patterns |