Data Analyst Portfolio Projects Using Indian Datasets

Two candidates apply for the same analyst role in Hyderabad. Both list "SQL, Power BI, Python." One has a GitHub link with three real projects on datasets the recruiter instantly understands — kirana sales, UPI payments, cab demand. Guess who gets the call. A portfolio is the single best way to stand out in India's crowded analyst market, and using relatable Indian datasets makes your work memorable.

Below are three end-to-end projects. For each you get the business question, the data, the steps, and the insights to showcase. Build all three and you have a complete, hireable portfolio.

Why Indian datasets?

• Relatability — a hiring manager grasps "weekend surge in UPI food orders" faster than a generic Kaggle dataset.
• Storytelling — local context lets you frame real business decisions.
• Differentiation — most applicants reuse the same Titanic/Iris datasets. You won't.

You can generate realistic synthetic data with Python (faker, numpy) or adapt public sources like data.gov.in, RBI/NPCI summaries, and open municipal data.

Project 1 — Kirana Store Sales Analysis (Excel + SQL + Power BI)

Business question: Which products, days and outlets drive revenue, and where is the kirana chain leaking margin?

The data: a sales table — date, outlet_id, city, product, category, quantity, unit_price, cost.

Steps:

1. Clean in Power Query: fix date formats, remove duplicate bills, handle negative quantities (returns).
2. Model: build a star schema — Sales fact + Date, Product, Outlet dimensions.
3. SQL analysis to answer specific questions:

SELECT city, category,
       SUM(quantity * unit_price)         AS revenue,
       SUM(quantity * (unit_price - cost)) AS margin
FROM sales
GROUP BY city, category
ORDER BY margin DESC;

4. DAX measures: Total Sales, Gross Margin %, Month-over-Month growth, Top-N products.
5. Dashboard: revenue by outlet, category mix, weekday vs. weekend patterns, slow-moving stock.

Insights to highlight: "Top 20% of SKUs drove 68% of revenue; staples had high volume but thin margin while packaged snacks carried the margin — recommend shelf-space rebalancing."

Project 2 — UPI Transactions Trends (SQL + Python)

Business question: When and where do people transact most, and which merchant categories dominate?

The data: a transactions table — txn_id, timestamp, user_id, merchant, category, amount, city, status.

Steps:

1. Load into Python with pandas; parse timestamps to extract hour, day-of-week, month.
2. Clean: drop failed/duplicate transactions, flag outlier amounts.
3. SQL/pandas analysis:

SELECT category,
       COUNT(*)        AS txns,
       ROUND(AVG(amount),0) AS avg_ticket
FROM transactions
WHERE status = 'SUCCESS'
GROUP BY category
ORDER BY txns DESC;

4. Visualize with matplotlib/seaborn: hourly transaction heatmap, category share, success-rate trend.
5. Optional: a simple cohort of repeat users.

Insights to highlight: "Transactions peak at 8-9 PM; food-delivery payments surge 22% on weekends; failure rate spikes during month-end salary credit — a reliability concern worth flagging."

Project 3 — Cab Demand Analysis (Python + Power BI)

Business question: Where and when is ride demand highest, and how should driver supply be positioned?

The data: a rides table — ride_id, request_time, pickup_zone, drop_zone, city, fare, wait_time, cancelled.

Steps:

1. EDA in Python: demand by hour, zone and day; cancellation patterns.
2. Feature work: peak vs. off-peak flag, surge windows, zone-level demand density.
3. Power BI dashboard: a map/matrix of demand by zone and hour, average wait time, cancellation hotspots, fare distribution.
4. Add a what-if slicer (e.g., simulate adding drivers in a high-cancellation zone).

Insights to highlight: "Morning 9-11 AM office-corridor demand outstrips supply, driving a 31% cancellation rate in two zones; reallocating drivers there could recover an estimated ₹X in lost rides."

How to present your portfolio

• GitHub repo per project with a clear README: business question → data → approach → key insight.
• Publish dashboards to the Power BI Service (or share PDF/screenshots).
• Write a short LinkedIn post per project — visibility brings recruiters.
• Link everything from your resume header.

What makes a project "interview-ready"

• It answers a business question, not just "I made charts."
• It shows the full pipeline: clean → model → analyze → visualize → recommend.
• It ends with a decision or recommendation, with a number attached.
• You can explain it in 60 seconds and defend your choices.

Three solid projects beat ten half-finished ones. Pick these, build them properly, and you'll have a portfolio that does the talking before you even enter the room.

Related: Power BI Portfolio Projects That Get You Hired · Build your skill score

Don't just read. Prove your skill on DevWithData.