TCS Power BI Interview Questions

CALCULATE modifies the filter context in which an expression is evaluated — the most powerful DAX function. FILTER returns a table satisfying a condition but does not change the context. CALCULATE replaces the existing filter on a column; FILTER creates a subset table. Pattern at TCS: CALCULATE(SUM(Sales[Amount]), FILTER(ALL(Sales), Sales[Region] = 'North')).

Define roles in Modeling > Manage Roles. DAX rule for relationship manager seeing only their accounts: [RM_ID] = USERPRINCIPALNAME(). Publish report and assign users to roles via Power BI Service > Security. For dynamic RLS at scale, maintain a security mapping table linking emails to branch IDs and use USERNAME() in a DAX relationship filter.

Star schema places the fact table at center with denormalized dimension tables — optimal for Power BI because the Vertipaq engine loves fewer joins. Snowflake schema normalizes dimensions, reducing storage but adding join complexity. Always recommend star schema for Power BI semantic models. Exception: when dimension tables exceed 10M rows.

High DAX query time: replace SUMX iterator over large tables with pre-aggregated measures. Too many visuals per page: paginate or use bookmarks. DirectQuery latency: switch non-volatile tables to Import mode. High-cardinality columns like GUIDs in relationships kill Vertipaq compression — encode those as integers. Check Aggregations for large fact tables.

Requires continuous date dimension marked as a Date Table. TOTALYTD gives year-to-date totals. SAMEPERIODLASTYEAR enables YoY comparison. DATEADD shifts context one month back for MoM analysis. At TCS, build a standard CalendarTable using CALENDAR() with Week, Month, Quarter, Fiscal Year, and IsWeekday columns — one table powers all time intelligence across projects.

Calculated Column: computed row-by-row during data refresh, stored in model — consumes RAM. Measure: computed dynamically at query time in current filter context — stores no data. Rule: use Calculated Columns only when you need to filter or slice by that value (like Age Group bucket). Use Measures for all aggregations shown in visuals. Never write SUM() as a Calculated Column — it wastes memory.