Tableau Last-Minute Revision

30 quick revision notes for Tableau

1. Dimensions are qualitative fields that slice data, while measures are numeric fields that Tableau aggregates, usually with SUM by default.
2. Blue pills are discrete and create headers; green pills are continuous and create axes, for both dimensions and measures.
3. A .twb file stores only the workbook XML, while a .twbx packages the workbook with extracts and local files.
4. Live connections query the source every time, while extracts are Hyper snapshots that are faster but need scheduled refreshes.
5. Relationships keep each table at its own grain and build joins at query time, avoiding duplicated measures from fixed joins.
6. Joins merge tables at row level in the physical layer and can inflate totals when the tables have different grains.
7. Data blending aggregates the secondary source to linking fields and behaves like a left join from the primary source.
8. Unions stack tables with the same columns vertically, and wildcard unions pick up new files automatically on refresh.
9. Order of operations: extract, data source, context filters, then Top N and FIXED, dimension filters, INCLUDE and EXCLUDE, measure filters, table calcs.
10. FIXED LOD ignores normal dimension filters; add the filter to context to make it change the FIXED result.
11. INCLUDE adds dimensions to the view level before re-aggregating, while EXCLUDE removes dimensions from the view level.
12. {FIXED : SUM([Sales])} returns the table-level grand total, useful for percent of total; wrap it in MIN, not SUM.
13. Build ratios as SUM([Profit]) / SUM([Sales]) so they are weighted correctly at every level of the view.
14. The cannot mix aggregate and non-aggregate error is fixed by aggregating all parts or using an LOD for row-level values.
15. IIF takes a test, true value, false value and optional unknown value; CASE matches exact values; IF handles ranges.
16. ZN converts null numbers to zero, IFNULL replaces nulls with a value, and ISNULL tests whether a value is null.
17. DATETRUNC returns a truncated date, DATEPART returns an integer part, and DATENAME returns the text name of a date part.
18. Table calculations run on aggregated results in the view, and their direction is set by addressing and partitioning fields.
19. In Compute Using, checked dimensions are addressing fields and unchecked dimensions become partitions where the calc restarts.
20. Use Specific Dimensions in table calculations so results stay correct when the view layout changes.
21. RANK gives 1,2,2,4 for ties, RANK_DENSE gives 1,2,2,3, and RANK_UNIQUE gives 1,2,3,4.
22. Table calculation filters such as LAST() < 6 hide marks after computing, so running totals stay correct.
23. Context filters run before Top N, so add a region filter to context to get the top products within that region.
24. Parameters are single-value variables that work only when referenced in calculations, filters, reference lines, bins or Top N.
25. Dynamic parameters can refresh their list of values and current value from data when the workbook opens.
26. Set actions change set membership on click, enabling drill downs and comparisons without removing other data.
27. Row-level security uses an entitlement table with a data source filter such as USERNAME() = [Username].
28. Site roles Creator, Explorer and Viewer set maximum capabilities, while project and content permissions refine access.
29. Published and certified data sources give one version of truth, central refresh schedules and shared calculations.
30. Improve performance with extracts, fewer marks and quick filters, simple calculations, fixed dashboard size and performance recording.

Night-before checklist

• Revise your saved questions on this page once, aloud.
• Read the quick notes and mark any you cannot explain in one line.
• Rehearse a two-minute self-introduction and two project stories.
• Keep CV, ID, laptop charger and interview link ready; sleep on time.