Katkat
Back

Compare Tableau blending, joins, and filters

TCSData Scientist · Technical Screen
HardData ScientistTechnical ScreenAnalytics & Experimentation

Last updated

This question evaluates a candidate's competency in Tableau data integration and visualization, specifically assessing understanding of joins versus blends, filter order of operations, aggregation and null-handling behavior, level-of-detail impacts, and cross-database/calculation limitations.

You have two datasets to analyze in Tableau.

Primary data source: 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 data source: Targets

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

Answer the following precisely:

  1. If you perform a physical inner join on State and Category, which rows remain and what are the values of SUM(Sales) and SUM(MonthlyTarget) by State, Category? Repeat for a left join (Orders → Targets).
  2. If instead you blend the Targets source into Orders (Orders = primary), explain exactly how Tableau computes the view when you place State and Category on rows and SUM(Sales), SUM(MonthlyTarget) on columns. Specify how unmatched members (e.g., NY–Technology; TX–Furniture; FL–Technology) appear, and whether their targets/sales show as null/zero. Why?
  3. Show how the results change under each of these filters and explain why based on Tableau’s order of operations: (a) a dimension filter Category = 'Furniture' from the Orders source; (b) a dimension filter Category = 'Furniture' from the Targets source; (c) a context filter State = 'CA' on Orders; (d) a measure filter SUM(Sales) > 150 applied to the view; (e) a data source filter on Targets keeping only Category IN ('Furniture','Technology').
  4. Describe two pitfalls where blending and joining yield different aggregates due to level of detail and many-to-many relationships (e.g., duplicate State–Category rows in Targets). Provide a concrete example from the given data.
  5. When would you prefer blending over joining in Tableau? Discuss cross-database analysis, required granularity, the linking field behavior, impacts on performance/caching, and limitations with calculations (e.g., FIXED LODs, table calcs) in secondary sources.