Tableau Data Integration: Joins, Blends, Filters, and Order of Operations

This exercise explores various methods of integrating and filtering data within Tableau, focusing on the distinctions between physical joins, data blending, and the impact of different filter types.

Data Sources

YouYou will work with two data sources:

• Primary (Orders)

  • Columns: OrderID, State, Category, Sales
  • Rows:
    1. (O1, CA, Furniture, 100)
    2. (O2, CA, Office Supplies, 200)
    3. (O3, NY, Technology, 150)
    4. (O4, TX, Furniture, 300)
    5. (O5, CA, Technology, 50)

• Secondary (Targets)

  • Columns: State, Category, MonthlyTarget
  • Rows: A) (CA, Furniture, 500) B) (CA, Technology, 400) C) (NY, Furniture, 250) D) (FL, Technology, 600)

Tasks

1. Physical Joins

   • Perform a physical Inner Join between Orders and Targets on State AND Category.
     • Identify which rows remain in the joined dataset.
     • Calculate SUM(Sales) and SUM(MonthlyTarget) by State and Category.
   • Perform a physical Left Join (with Orders as the left table) between Orders and Targets on State AND Category.
     • Identify which rows remain in the joined dataset.
     • Calculate SUM(Sales) and SUM(MonthlyTarget) by State and Category.

2. Data Blending

   • Configure data blending using Orders as the Primary data source and Targets as the Secondary data source. Link the sources on State and Category.
   • With State and Category on Rows, and SUM(Sales) and SUM(MonthlyTarget) on Columns, explain precisely how Tableau computes the resulting view.
   • Specify how unmatched members (e.g., NY–Technology; TX–Furniture; FL–Technology) appear in the view, indicating whether their corresponding sales/targets show as NULL or zero. Provide a detailed explanation for this behavior.

3. Filter Impact and Order of Operations

   • Describe how the results from the blending exercise (Task 2) would change under each of the following filter scenarios. For each, explain why the change occurs, referencing Tableau’s order of operations:
     • (a) A Dimension filter applied to Category = 'Furniture' from the Orders (primary) data source.
     • (b) A Dimension filter applied to Category = 'Furniture' from the Targets (secondary) data source.
     • (c) A Context filter applied to State = 'CA' on the Orders data source.
     • (d) A Measure filter applied to SUM(Sales) > 150.
     • (e) A Data source filter applied to Targets keeping only Category IN ('Furniture', 'Technology').

4. Blending vs. Joining Pitfalls

   • Describe two common pitfalls where data blending and physical joining can yield different aggregate results due to differences in level of detail or the presence of many-to-many relationships (e.g., duplicate State–Category rows in Targets).
   • Provide a concrete example based on the given Orders and Targets data to illustrate one of these pitfalls.

5. When to Prefer Blending

   • Discuss scenarios where you would prefer using data blending over physical joining in Tableau. Your answer should address the following considerations:
     • Cross-database analysis capabilities.
     • Required granularity of data.
     • Behavior of linking fields.
     • Performance and caching implications.
     • Limitations on calculations (e.g., FIXED Level of Detail expressions, table calculations) when using secondary sources.