Day 28: Power BI — Building Real Projects

Recommended datasets from Kaggle: (1) Superstore Sales Dataset — classic retail dataset with orders, regions, categories. Build a complete sales dashboard with YoY, top products, regional maps. (2) HR Analytics Dataset — employee attrition with demographics, satisfaction scores, tenure. Build an attrition dashboard. (3) COVID-19 dataset — global or India-specific. Build a tracking dashboard with time intelligence. (4) IPL Dataset — cricket statistics. Build a player/team performance dashboard. (5) Indian Startup Funding dataset — build a funding trend analysis. Choose a dataset from a domain the interviewer works in — it shows targeted preparation.

A complete portfolio project should demonstrate: (1) Proper data model — star schema with fact and dimension tables, clearly named. (2) Power Query transformations — at least data type changes, null handling, and one merge or append. (3) Date table — created in DAX or Power Query, marked as Date table. (4) Core measures — Total Revenue, YoY comparison, Achievement %, using CALCULATE and SAMEPERIODLASTYEAR. (5) Slicers for filtering — at minimum Year and one dimension. (6) Multiple visual types — at least a KPI card, bar chart, line chart, and table/matrix. (7) Conditional formatting in the table. (8) Clean layout — white space, consistent colours, clear titles. Host on Power BI Service and share the link.

Use the impact-first format: "Built an end-to-end sales analytics dashboard in Power BI using a 50K-row retail dataset — including a star schema data model, 15 DAX measures (YoY growth, running totals, RANKX for top products), and an interactive 4-page report with drill-through and RLS. Identified that the Technology category had 2x higher profit margin than Furniture despite similar revenue — a finding actionable for product mix strategy." Key elements: size of data, specific DAX functions used, number of pages/visuals, and a concrete insight or outcome. Never just write "built a Power BI dashboard" — always add specifics.

Practical tests typically give you a dataset (Excel or CSV) and ask you to: connect and clean the data, build a data model, create 3-5 measures, and build a 1-2 page report in 30-60 minutes. Preparation: (1) Practice the end-to-end workflow on unfamiliar datasets — not just learning tools in isolation. (2) Build a personal checklist: Table → Power Query clean → Date table → relationships → base measures → visuals → formatting. (3) Speed matters — know keyboard shortcuts, know which field goes where for common visuals without thinking. (4) Explain your thinking out loud — interviewers give credit for your approach even if the output is not perfect. (5) Handle missing requirements by making a reasonable assumption and stating it.

(1) Dynamic titles: chart title changes based on slicer selection using SELECTEDVALUE — shows attention to user experience. (2) Tooltip pages: hover over a bar to see a detailed breakdown — shows advanced feature knowledge. (3) Bookmark navigation: buttons navigate between views without page changes — shows professional UI design. (4) A clear business narrative: a text box on the first page explains "this report answers: which regions are above target and why?" — shows business thinking, not just technical. (5) Documented data model: a page in the report (or external README) explains the data sources, relationships, and key measure logic — shows professional standards and makes the project credible.