Tableau = visual analytics platform for business intelligence
Makes data exploration and visualization accessible to everyone

Tableau Product Suite:
┌─────────────────────┬──────────────────────────────┐
│ Product             │ Purpose                       │
├─────────────────────┼──────────────────────────────┤
│ Tableau Desktop     │ Authoring tool (create viz)   │
│                     │ Connect, analyze, build        │
├─────────────────────┼──────────────────────────────┤
│ Tableau Server      │ On-premise sharing platform   │
│                     │ Publish, collaborate, govern   │
├─────────────────────┼──────────────────────────────┤
│ Tableau Cloud       │ SaaS version of Server        │
│ (formerly Online)   │ No infrastructure to manage   │
├─────────────────────┼──────────────────────────────┤
│ Tableau Prep        │ Data preparation & cleaning   │
│                     │ Visual ETL tool                │
├─────────────────────┼──────────────────────────────┤
│ Tableau Public      │ Free version for public data  │
│                     │ Portfolios, learning           │
├─────────────────────┼──────────────────────────────┤
│ Tableau Reader      │ Free viewer for .twbx files   │
│                     │ View only, no editing          │
└─────────────────────┴──────────────────────────────┘

Workflow: Desktop (create) → Server/Cloud (share)
File types: .twb (workbook), .twbx (packaged with data),
            .tds (data source), .hyper (extract)

Dimensions (Blue pills):
- Categorical / qualitative data
- Names, dates, categories, IDs
- Used to SLICE and GROUP data
- Examples: Product Name, Region, Order Date, Category

Measures (Green pills):
- Quantitative / numerical data
- Values that can be aggregated
- SUM, AVG, COUNT, MIN, MAX
- Examples: Sales, Profit, Quantity, Discount

Visual difference in Tableau:
┌──────────────────────────────────────┐
│  Data Pane                           │
│  ┌─────────────────────────────────┐ │
│  │ 🔵 Dimensions                   │ │
│  │   Category                      │ │
│  │   Customer Name                 │ │
│  │   Order Date                    │ │
│  │   Region                        │ │
│  ├─────────────────────────────────┤ │
│  │ 🟢 Measures                     │ │
│  │   Sales                         │ │
│  │   Profit                        │ │
│  │   Quantity                      │ │
│  │   Discount                      │ │
│  └─────────────────────────────────┘ │
└──────────────────────────────────────┘

Key insight:
- Dimensions SLICE data (create categories/groups)
- Measures AGGREGATE data (calculate values)
- Drag Region (dimension) → creates one bar per region
- Drag Sales (measure) → shows SUM of sales per region

You CAN convert between them:
- Right-click a measure → "Convert to Dimension"
- Useful for: ZIP codes, IDs stored as numbers

Live Connection:
- Queries database in REAL-TIME
- Always shows current data
- Performance depends on database speed
- Requires active database connection
- Best for: real-time dashboards, small datasets

Extract (.hyper file):
- SNAPSHOT of data stored locally
- Faster performance (columnar storage)
- Works offline
- Needs scheduled refresh for updates
- Best for: large datasets, complex calculations

Comparison:
┌──────────────┬──────────────┬──────────────┐
│ Feature      │ Live         │ Extract      │
├──────────────┼──────────────┼──────────────┤
│ Speed        │ Slower       │ Faster       │
│ Data         │ Real-time    │ Snapshot     │
│ Offline      │ No           │ Yes          │
│ File size    │ No local     │ .hyper file  │
│ Refresh      │ Automatic    │ Scheduled    │
│ Complex calc │ Slower       │ Much faster  │
│ Row-level    │ Yes          │ At refresh   │
│ security     │              │              │
└──────────────┴──────────────┴──────────────┘

When to use Live:
- Real-time monitoring dashboards
- Small datasets (< 1M rows)
- Data changes frequently and must be current

When to use Extract:
- Large datasets (millions of rows)
- Complex LOD expressions
- Users need offline access
- Database is slow or has limited connections

Calculated fields = custom calculations using
Tableau functions (like Excel formulas but more powerful)

// Profit Ratio
SUM([Profit]) / SUM([Sales])

// Running Total
RUNNING_SUM(SUM([Sales]))

// Date Difference (shipping days)
DATEDIFF('day', [Order Date], [Ship Date])

// Conditional calculation
IF [Profit] > 0 THEN "Profitable"
ELSEIF [Profit] = 0 THEN "Break Even"
ELSE "Loss"
END

// String manipulation
UPPER(LEFT([Customer Name], 1)) +
  LOWER(MID([Customer Name], 2))

// Year-over-Year Growth
(SUM([Sales]) - LOOKUP(SUM([Sales]), -1))
  / ABS(LOOKUP(SUM([Sales]), -1))

// Common function categories:
// Number:  ABS, ROUND, CEILING, FLOOR, ZN
// String:  LEFT, RIGHT, MID, CONTAINS, REPLACE
// Date:    DATEADD, DATEDIFF, DATEPART, TODAY
// Logic:   IF/THEN, CASE, IIF, ZN, IFNULL
// Table:   RUNNING_SUM, RANK, WINDOW_AVG
// Aggregate: SUM, AVG, COUNT, COUNTD, MIN, MAX

// ZN() is Tableau-specific:
// ZN([Sales]) → replaces NULL with 0
// Very useful for avoiding NULL errors

LOD expressions = compute aggregations at a DIFFERENT
level of detail than what's shown in the view

Three types: FIXED, INCLUDE, EXCLUDE

1. FIXED — compute at specified dimension(s)
   Ignores view filters and dimensions

   // Total sales per customer (regardless of view)
   {FIXED [Customer Name] : SUM([Sales])}

   // Use case: Show customer total even when
   // view is broken down by product category

2. INCLUDE — adds a dimension to the calculation
   More granular than the view

   // Average sales per order, then aggregate
   {INCLUDE [Order ID] : SUM([Sales])}
   // Then AVG() this in the view

3. EXCLUDE — removes a dimension from calculation
   Less granular than the view

   // Total sales excluding the month dimension
   {EXCLUDE [Order Date] : SUM([Sales])}
   // Shows yearly total even in monthly view

Real interview example:
"Show each customer's first purchase date alongside
their current orders"

{FIXED [Customer Name] : MIN([Order Date])}

// This returns the FIRST order date for each customer
// regardless of what filters or dimensions are in view

Order of operations:
FIXED → computed first (before dimension filters)
INCLUDE/EXCLUDE → computed after dimensions
Context filters → applied before FIXED

Joins (Data Source Level):
- Combine tables from the SAME database/connection
- Happens at the data source level
- Types: Inner, Left, Right, Full Outer
- Creates a single combined dataset
- Better performance for same-database tables

Data Blending (View Level):
- Combine data from DIFFERENT sources
- Primary source + Secondary source(s)
- Happens at the visualization level
- Uses LEFT JOIN logic on linking fields
- Secondary source shows * (aggregated) values

Comparison:
┌──────────────┬──────────────────┬──────────────────┐
│ Feature      │ Joins            │ Data Blending    │
├──────────────┼──────────────────┼──────────────────┤
│ Sources      │ Same database    │ Different sources│
│ Level        │ Data source      │ View/sheet       │
│ Performance  │ Better           │ Slower           │
│ Granularity  │ Row-level        │ Aggregated       │
│ Join types   │ All (inner,left) │ Left join only   │
│ Setup        │ Data pane        │ Automatic/manual │
└──────────────┴──────────────────┴──────────────────┘

When to use Joins:
- Tables in the same database
- Need row-level detail
- Better performance needed

When to use Data Blending:
- Data from different sources (SQL + Excel + API)
- Different granularity levels
- Cannot modify the data source

// Tableau also has Relationships (newer):
// - Automatic join type selection
// - Context-aware (joins only when needed)
// - Recommended over traditional joins

Chart Selection Guide:

┌──────────────┬──────────────────────────────────┐
│ Chart Type   │ Best For                          │
├──────────────┼──────────────────────────────────┤
│ Bar Chart    │ Comparison across categories      │
│              │ "Sales by Region", "Top Products" │
├──────────────┼──────────────────────────────────┤
│ Line Chart   │ Trends over time                  │
│              │ "Monthly Revenue", "Growth Rate"  │
├──────────────┼──────────────────────────────────┤
│ Scatter Plot │ Correlation between 2 measures    │
│              │ "Sales vs Profit", "Age vs Income"│
├──────────────┼──────────────────────────────────┤
│ Map          │ Geographic data                   │
│              │ "Sales by State", "Store Locations"│
├──────────────┼──────────────────────────────────┤
│ Pie Chart    │ Part-to-whole (AVOID if > 5 cats) │
│              │ "Market Share" (use bar instead)  │
├──────────────┼──────────────────────────────────┤
│ Treemap      │ Hierarchical part-to-whole        │
│              │ "Revenue by Category > Product"   │
├──────────────┼──────────────────────────────────┤
│ Heatmap      │ Density and patterns              │
│              │ "Sales by Day × Hour"             │
├──────────────┼──────────────────────────────────┤
│ Bullet Graph │ Actual vs Target                  │
│              │ "KPI Performance"                 │
├──────────────┼──────────────────────────────────┤
│ Gantt Chart  │ Duration and timelines            │
│              │ "Project Timeline"                │
└──────────────┴──────────────────────────────────┘

Golden rule: Bar charts answer 80% of business questions.
When in doubt, start with a bar chart.

Avoid: 3D charts, dual-axis with different scales,
pie charts with many slices, excessive color coding

Interactive Dashboard Components:

1. Filters
   - Quick filters (dropdown, slider, checkbox)
   - Context filters (improve performance)
   - Data source filters (pre-filter data)

2. Parameters
   - User-controlled variables
   - Dropdown, slider, type-in
   - Drive calculations and filters dynamically

3. Actions (key for interactivity)
   ┌──────────────┬──────────────────────────┐
   │ Filter Action│ Click chart A → filters B│
   │ Highlight    │ Hover on A → highlights B│
   │ URL Action   │ Click → opens webpage    │
   │ Set Action   │ Click → updates a set    │
   │ Parameter    │ Click → updates parameter│
   └──────────────┴──────────────────────────┘

4. Dashboard Layout
   - Fixed size: pixel-perfect, specific screen
   - Automatic: responsive, adapts to screen
   - Range: min/max dimensions

Best Practices:
- 3-5 visualizations max per dashboard
- Clear, descriptive title
- Consistent color palette (max 5-7 colors)
- Filter actions for drill-down (click to filter)
- KPI summary at top (BANs - Big Ass Numbers)
- Left-to-right, top-to-bottom reading flow
- Mobile layout for phone/tablet users
- Tooltips with additional context

// Interview tip: Always mention "filter actions"
// as the primary interactivity mechanism.
// "Click on a region in the map to filter
// all other charts to that region."

Performance Optimization Checklist:

1. Data Source Level:
   ✓ Use extracts over live connections
   ✓ Aggregate data at the source (SQL)
   ✓ Hide unused fields from data source
   ✓ Use custom SQL only when necessary
   ✓ Filter data at source level

2. Calculation Level:
   ✓ Avoid COUNTD (expensive operation)
   ✓ Minimize LOD expressions when possible
   ✓ Use FIXED LOD instead of INCLUDE/EXCLUDE
   ✓ Boolean calculations > IF/THEN
   ✓ Move calculations to database (SQL)

3. Visualization Level:
   ✓ Reduce number of marks (< 10,000 ideal)
   ✓ Limit filters (use context filters)
   ✓ Avoid "Show All Values" on filters
   ✓ Use filter actions instead of quick filters
   ✓ Minimize dashboard objects

4. Context Filters:
   - Regular filter: applied independently
   - Context filter: creates a temporary table
   - Other filters run AGAINST the context
   - Use when: one filter dramatically reduces data

5. Performance Recording:
   Help → Settings → Start Performance Recording
   → Identifies slow queries and calculations
   → Shows time per query, layout, rendering

Common performance killers:
✗ COUNTD on large datasets
✗ Too many LOD expressions
✗ Live connection to slow database
✗ 100+ marks on a single chart
✗ Multiple quick filters with "All Values"

Parameters:
- User-controlled variables (like a form input)
- Types: Integer, Float, String, Boolean, Date
- Display as: Dropdown, Slider, Type-in
- Drive calculations, filters, reference lines

// Example: Top N parameter
// 1. Create parameter "Top N" (integer, range 1-20)
// 2. Create calculated field:
//    RANK(SUM([Sales])) <= [Top N Parameter]
// 3. Use as filter → user controls how many to show

// Example: Metric selector
// 1. Create parameter "Select Metric" (string)
//    Values: "Sales", "Profit", "Quantity"
// 2. Create calculated field:
//    CASE [Select Metric]
//      WHEN "Sales" THEN SUM([Sales])
//      WHEN "Profit" THEN SUM([Profit])
//      WHEN "Quantity" THEN SUM([Quantity])
//    END

Sets:
- Custom subsets of data based on conditions
- Two types:
  ┌──────────────┬──────────────────────────┐
  │ Fixed Set    │ Manually selected members│
  │              │ Does NOT change           │
  ├──────────────┼──────────────────────────┤
  │ Dynamic Set  │ Based on a condition     │
  │              │ Updates automatically     │
  └──────────────┴──────────────────────────┘

// Dynamic set example:
// "Top 10 Customers by Sales"
// → Automatically updates as data changes

// Combined: Parameter-driven sets
// User selects threshold via parameter
// Set updates based on parameter value
// → Fully interactive analysis