📘 DATA ANALYTICS SERIES · CHAPTER 59

Data Analyst Portfolio Projects — India 2026

10 project ideas with datasets, tools, step-by-step build guides, and exactly what to say about each project in an interview — designed for the Indian job market, from beginner to advanced.

⏱ 22 min read📅 September 2026📍 India / Delhi NCR

Why Portfolio Projects Matter More Than Certifications

In India's data analytics job market in 2026, a portfolio of 3–5 strong projects consistently outperforms a wall of certifications in hiring decisions. Certifications prove you paid for and passed a test. Projects prove you can analyse data, draw conclusions, and communicate findings — which is the actual job.

This guide gives you 10 projects across SQL, Python, Power BI, and Excel, with Indian-context datasets, step-by-step build instructions, and the exact framing to use when discussing each project in interviews.

Difficulty:BeginnerIntermediateAdvancedBuild 3–5 projects; aim for at least one Intermediate or Advanced
1

Indian E-Commerce Sales Dashboard

BeginnerRetail / E-CommercePower BIExcel
Dataset
Kaggle: "E-Commerce Sales Dataset India" — orders, products, regions, returns
Deliverable
Power BI dashboard with revenue by category, region map, monthly trend, top 10 products, return rate card
Build Steps
  1. Load CSV into Power BI via Power Query
  2. Create date table and mark as Date table
  3. Build measures: Total Revenue, MoM Growth %, Return Rate, Avg Order Value
  4. Design 4-page report: Executive Summary, Product Analysis, Regional Map, Returns
  5. Add slicers: category, state, date range
💬 How to talk about this in an interview
"I built this to simulate the reporting a business analyst at an Indian e-commerce company would produce. The insight that surprised me was that the South region had 23% higher return rates than North — the dashboard makes that visible with one click."
2

SQL Customer Churn Analysis

BeginnerTelecom / SubscriptionSQL (PostgreSQL)Excel / Python for visuals
Dataset
Kaggle: "Telco Customer Churn" — adapt with Indian city names in Excel before loading
Deliverable
SQL query file (.sql) on GitHub + a slide with 3 findings and recommendations
Build Steps
  1. Load data into PostgreSQL (free local install)
  2. Define churn: customers with no activity in last 90 days
  3. Segment: churn rate by tenure band, plan type, contract length
  4. Find: which combination of tenure + plan has highest churn?
  5. Write a 1-page recommendation: which segment to target for retention first
-- Churn rate by tenure band and contract type
SELECT
    CASE
        WHEN tenure_months <= 3  THEN '0–3 months'
        WHEN tenure_months <= 12 THEN '4–12 months'
        ELSE '12+ months'
    END AS tenure_band,
    contract_type,
    COUNT(*)                                          AS total_customers,
    SUM(CASE WHEN churned = 1 THEN 1 ELSE 0 END)     AS churned,
    ROUND(100.0 * SUM(CASE WHEN churned = 1 THEN 1 ELSE 0 END)
          / COUNT(*), 1)                              AS churn_pct
FROM customers
GROUP BY tenure_band, contract_type
ORDER BY churn_pct DESC;
💬 How to talk about this in an interview
"This project shows how I think about business problems in SQL. The key finding was that month-to-month contract users in their first 3 months had 3× the churn rate of annual contract users — recommending a 3-month free upgrade to annual as a retention trigger."
3

Python EDA — Indian IPL / Cricket Dataset

BeginnerSports AnalyticsPython (pandas, matplotlib, seaborn)Jupyter Notebook
Dataset
Kaggle: "IPL Complete Dataset" — matches, deliveries, players 2008–2024
Deliverable
Jupyter Notebook (.ipynb) on GitHub with clean markdown cells explaining each analysis step
Build Steps
  1. Load matches.csv and deliveries.csv
  2. Clean: handle missing values, fix team name inconsistencies over seasons
  3. Analyse: win rate by toss decision, home vs away performance, top run scorers by year
  4. Visualise: seaborn heatmap of toss-win vs match-win correlation by venue
  5. Write a conclusion section: "What this data tells us about toss strategy"
💬 How to talk about this in an interview
"I chose cricket because it's familiar to Indian interviewers and makes the business analogy clear. The toss-win correlation changes significantly by venue — which mirrors how context changes decision value in business analytics."
4

HR Analytics — Attrition Dashboard

IntermediateHR / People AnalyticsPower BIPython for preprocessing
Dataset
Kaggle: "IBM HR Analytics Employee Attrition" — commonly used, adapt job titles to Indian context
Deliverable
Power BI report with attrition KPIs, department breakdown, salary band analysis, and a "risk score" calculated column
Build Steps
  1. Preprocess in Python: encode categoricals, create AttritionRisk score using rule-based logic
  2. Load into Power BI
  3. Build: overall attrition %, by department, by salary band, by tenure group
  4. Create conditional formatting heatmap: department vs tenure attrition rate
  5. Add a "Top 5 Risk Factors" text card using DAX CONCATENATEX
💬 How to talk about this in an interview
"HR attrition analysis is a common real-world brief for data analysts in Indian companies. I framed it as: if I were presenting to the CHRO, what three decisions would this dashboard inform? That's the lens I used to choose which visuals to include."
5

RBI / SEBI Financial Data Analysis

IntermediateFinance / BankingPython (pandas, matplotlib)SQL for aggregations
Dataset
RBI Handbook of Statistics on Indian Economy (data.rbi.org.in) — free, official, credible
Deliverable
Python notebook analysing 10 years of bank credit growth, NPA trends, or repo rate vs inflation correlation
Build Steps
  1. Download CSV data from RBI's DBIE portal
  2. Clean with pandas: handle Indian number formatting (lakh/crore), parse fiscal years
  3. Analyse: repo rate changes vs retail inflation lag (typically 2–3 quarter lag)
  4. Plot: dual-axis line chart of repo rate and CPI inflation 2014–2024
  5. Write a 200-word interpretation section
💬 How to talk about this in an interview
"Using RBI official data shows interviewers you can work with real public-sector data formats, not just clean Kaggle CSVs. The data cleaning step here was non-trivial — which is exactly what real analyst work looks like."
6

Customer Segmentation with K-Means

IntermediateRetail / CRMPython (pandas, sklearn, seaborn)
Dataset
Kaggle: "Online Retail II" dataset — ~1M rows of UK transactions, adapt product names to Indian context
Deliverable
Python notebook + a one-page segment profile: what each cluster looks like and what action it triggers
Build Steps
  1. Calculate RFM (Recency, Frequency, Monetary) per customer
  2. Scale with StandardScaler
  3. Elbow method to choose K (typically 4–5 for retail)
  4. Profile each cluster: mean RFM values, size, revenue share
  5. Name and describe each segment: Champions, Loyal, At Risk, Lost
💬 How to talk about this in an interview
"This project shows I can do unsupervised ML, but more importantly it shows I can translate clusters into business actions — Champions get a loyalty programme, At Risk get a win-back offer with a discount. The analysis is only useful if it drives a decision."
7

SQL Window Functions — Sales Ranking Dashboard

IntermediateRetail / SalesSQL (PostgreSQL or BigQuery)Tableau Public or Power BI for visuals
Dataset
Kaggle: "Superstore Sales Dataset" — US data, easy to rename to Indian cities/regions
Deliverable
SQL file with 5+ advanced queries + exported results visualised in Tableau Public
Build Steps
  1. Load into PostgreSQL
  2. Write: RANK() of products by revenue per region per quarter
  3. Write: LAG() to calculate MoM revenue change per category
  4. Write: running total of revenue per salesperson using SUM() OVER()
  5. Write: NTILE(4) to quartile customers by spend
  6. Publish results in Tableau Public as an interactive dashboard
-- Monthly revenue with MoM change using LAG
SELECT
    category,
    TO_CHAR(order_date, 'YYYY-MM')  AS month,
    SUM(sales)                       AS monthly_revenue,
    LAG(SUM(sales)) OVER (
        PARTITION BY category
        ORDER BY TO_CHAR(order_date, 'YYYY-MM')
    )                                AS prev_month_revenue,
    ROUND(100.0 * (SUM(sales) - LAG(SUM(sales)) OVER (
        PARTITION BY category
        ORDER BY TO_CHAR(order_date, 'YYYY-MM')
    )) / NULLIF(LAG(SUM(sales)) OVER (
        PARTITION BY category
        ORDER BY TO_CHAR(order_date, 'YYYY-MM')
    ), 0), 1)                        AS mom_change_pct
FROM orders
GROUP BY category, TO_CHAR(order_date, 'YYYY-MM')
ORDER BY category, month;
💬 How to talk about this in an interview
"This is designed specifically to show SQL window function depth — because that's what most SQL interview rounds test. I can walk through any of these queries and explain why I chose RANK vs DENSE_RANK for the product ranking."
8

A/B Test Analysis — Python Statistics

IntermediateProduct / GrowthPython (pandas, scipy, statsmodels)
Dataset
Kaggle: "A/B Testing Dataset" — website conversion data
Deliverable
Python notebook with full A/B test analysis: sample size check, z-test, confidence interval, business recommendation
Build Steps
  1. Load data, split into control and treatment groups
  2. Check: was the test run long enough? (check sample size vs required)
  3. Calculate conversion rates for both groups
  4. Run two-proportion z-test with scipy.stats
  5. Calculate 95% confidence interval for the lift
  6. Write recommendation: ship or don't ship, with reasoning
💬 How to talk about this in an interview
"Product analytics interviews always include A/B testing. This project shows I know the full workflow — not just running the t-test but checking sample size first and interpreting the confidence interval as a business range, not just a pass/fail."
9

Excel MIS Dashboard — Operational KPIs

BeginnerOperations / MISExcel (Power Query, Pivot Tables, XLOOKUP, charts)
Dataset
Create your own: a 500-row fictional Indian manufacturing or logistics dataset (ChatGPT can generate this)
Deliverable
Excel workbook with a 1-page dashboard: 6 KPI cards, 2 trend charts, 2 breakdowns, slicer for month
Build Steps
  1. Build raw data tab with realistic Indian business data
  2. Power Query: clean and load with a refresh-ready connection
  3. Pivot Table tab: aggregate KPIs by category and month
  4. Dashboard tab: use GETPIVOTDATA or INDEX-MATCH to pull KPI values into formatted cells
  5. Add slicers connected to all pivot tables
  6. Screenshot the final dashboard for LinkedIn
💬 How to talk about this in an interview
"I built this to practice the MIS analyst workflow that is common in Indian manufacturing and logistics companies. Every element is formula-driven — clicking the month slicer updates all 6 KPIs and both charts instantly. This is what a real Monday morning report looks like."
10

End-to-End Capstone — Indian Startup Metrics

AdvancedProduct / BusinessSQLPythonPower BI
Dataset
Build your own: simulate 12 months of fictional Indian app startup data (users, events, subscriptions) using Python Faker or ChatGPT
Deliverable
Full multi-tool project: SQL for data modelling → Python for EDA and churn model → Power BI for executive dashboard → a written "analyst memo" summarising findings
Build Steps
  1. Design the data model: users, events, subscriptions, payments tables
  2. Generate synthetic Indian data (names, cities, UPI/card payments)
  3. SQL: build user funnel analysis, retention cohorts, revenue queries
  4. Python: churn prediction using Logistic Regression (see Ch 57)
  5. Power BI: executive dashboard with north star metric prominently featured
  6. Analyst memo: 1-page written summary of findings and 3 recommendations
💬 How to talk about this in an interview
"This is the project I lead with in interviews. It shows SQL, Python, and Power BI working together end-to-end — which mirrors exactly how a real analyst at a startup uses all three tools in a single week. The written memo section shows I can translate data into decisions, not just visuals."

How to Structure and Host Your Portfolio

GitHub
SQL .sql files, Python .ipynb notebooks, README explaining the project, data cleaning notes, key findings summary
💡 Write a README.md for every project. Include: Problem Statement, Dataset, Tools Used, Key Findings, and a screenshot of the output.
Tableau Public
Published Power BI screenshots or Tableau interactive dashboards — shareable via URL at no cost
💡 Always embed a brief "what this shows" annotation on the dashboard. Interviewers may look at it without you present.
LinkedIn Featured
Pin your best project PDF, a post walking through your analysis, or a link to your GitHub or Tableau Public
💡 Write one LinkedIn post per project as you complete it. Describe the problem, what you found, and what you learned. 500+ word posts perform well in Indian analytics community.

Frequently Asked Questions

How many portfolio projects does a fresher data analyst need in India?

3–5 well-documented projects are better than 10 rushed ones. Quality beats quantity. Each project should show end-to-end work: data sourcing or loading, cleaning, analysis, visualisation, and a business recommendation. A portfolio with 3 strong projects covering SQL, Python, and Power BI will outperform a portfolio with 10 basic "visualise this CSV" projects.

What datasets should I use for a data analyst portfolio in India?

Best sources for Indian-context datasets: Kaggle (filter by "India"), data.gov.in (government open data), SEBI/NPCI published datasets, RBI macroeconomic data, and your own scraping of public platforms. Using Indian datasets (Indian e-commerce orders, Indian cricket statistics, Indian election data) makes your projects more relevant to Indian interviewers than generic US datasets.

Where should I host my data analyst portfolio in India?

Three options: (1) GitHub — host SQL scripts, Python notebooks (.ipynb), and a README that explains each project. Free, widely respected by technical interviewers. (2) Tableau Public — publish interactive dashboards for free. Shareable via URL. (3) LinkedIn Featured section — pin your best projects as PDFs, images, or external links. The best portfolios use all three and link between them.

How do I present my portfolio project in a data analyst interview?

Use the STAR format adapted for analysis: Situation (what business problem were you solving?), Task (what data did you have and what were you trying to find?), Analysis (what technique did you use and why?), Result (what did you find and what would you recommend?). Lead with the business finding, not the technical method. "I found that 40% of orders that cancelled came from first-time buyers during flash sales — suggesting the checkout flow was too complex for new users" is stronger than "I used pandas groupby with a pivot table."

Can I build a data analyst portfolio without work experience in India?

Yes — many Indian data analysts land their first role with a portfolio built entirely on public datasets and personal projects. The key is framing: present every project as if it were a real work assignment. Write a problem statement, define your stakeholder (e.g., "as an analyst for a fictional Indian e-commerce company"), do the analysis, and write a recommendation. The quality of your thinking and communication matters more than whether the data was from a real employer.

Build These Projects With Guidance — Not Alone

At Evika Academy, Noida Sector 51, students build guided portfolio projects on Indian datasets as part of the course — so every project is interview-ready before they graduate.

📱 Book Free Demo on WhatsApp
🎓 Free Demo Class — Online & Offline · Noida Sector 51