What a Data Analyst Portfolio Actually Needs
A hiring manager reviewing your portfolio is trying to answer one question: can this person take raw, messy data and turn it into something useful? Each project should demonstrate a complete cycle — data acquisition, cleaning, analysis, and a clear output. Here is what a strong portfolio covers:
At least 1 SQL project
Complex queries, joins, aggregations. Proves database competency.
At least 1 Python project
Pandas cleaning + analysis. Proves automation and scale.
At least 1 dashboard
Power BI or Tableau. Proves you can communicate findings visually.
Business framing
Every project has a business question, not just a data output.
Minimum viable portfolio: 3 projects covering SQL + Python + a dashboard. Four to five is stronger. More than six is unnecessary — depth beats breadth. Each project needs a README that explains the business question, your approach, and the finding.
Project 01Beginner⏱ 1–2 weeks · 🛠 SQL + Excel or Power BI
E-commerce Sales Analysis
DATASET
Kaggle: "Brazilian E-Commerce Public Dataset by Olist" (100k+ orders, free download)
BUSINESS QUESTION
"Which product categories drive the most revenue, and which have the highest return rates? How does delivery time affect customer review scores?"
DELIVERABLES
- SQL queries: revenue by category, top 10 products, monthly order trend
- Cleaned dataset (handle nulls in delivery dates, standardise city names)
- Power BI / Excel dashboard: revenue KPIs, category breakdown, monthly trend, delivery time vs rating scatter
WHAT IT PROVES TO A HIRING MANAGER · Basic SQL aggregations, data cleaning, and stakeholder-ready dashboard design. This is the entry-level project — do it well and thoroughly.
README TIP · State the business question first. Include 3 key findings (not just charts). Example: "Categories with >5 day delivery time have a 0.4 lower average rating — a potential ₹2.3M annual revenue risk."
Project 02Beginner–Intermediate⏱ 1–2 weeks · 🛠 Python (Pandas) + Matplotlib/Seaborn
HR Attrition Analysis
DATASET
IBM HR Analytics Employee Attrition Dataset — available on Kaggle (1470 rows, 35 columns, synthetic data)
BUSINESS QUESTION
"What factors most strongly predict employee attrition? Which department, role, and seniority combinations have the highest attrition risk?"
DELIVERABLES
- Python notebook: data cleaning, exploratory analysis, correlation with attrition
- At least 6 visualisations: attrition rate by department, by age group, by overtime status, by job satisfaction, by salary band, attrition rate over years at company
- Written summary: top 3 factors driving attrition, with a recommendation for each
WHAT IT PROVES TO A HIRING MANAGER · Pandas EDA, Matplotlib/Seaborn visualisation, ability to connect data to business action (not just charts). Also shows domain awareness in HR analytics.
README TIP · Frame findings as business recommendations. "Employees with overtime=Yes have 2.6× higher attrition. Recommendation: audit overtime distribution in Sales and R&D departments."
Project 03Intermediate⏱ 1–2 weeks · 🛠 SQL (PostgreSQL or SQLite)
SQL Sales Performance Report
DATASET
Northwind database (classic sample DB — available free as PostgreSQL or SQLite, covers orders, products, employees, customers)
BUSINESS QUESTION
"Which sales reps are over/under-performing versus target? What is the month-over-month trend for each category? Which customers are at churn risk (no order in 90+ days)?"
DELIVERABLES
- SQL script file with 8–10 queries: monthly revenue, rep performance ranking, category YoY growth, top customers by LTV, 90-day inactive customers
- Window functions used: RANK, LAG for MoM, SUM OVER for running totals
- CTEs used: at least one multi-step CTE for a complex calculation
- Results exported as CSV for visualisation
WHAT IT PROVES TO A HIRING MANAGER · Advanced SQL — window functions, CTEs, multi-table JOINs, date arithmetic. This is the project that most directly maps to day-one job requirements.
README TIP · Include the question above each query as a comment. Show the output (first 5 rows as a table in the README). State what business decision each query enables.
Project 04Intermediate⏱ 2 weeks · 🛠 Power BI + DAX
Power BI Financial Dashboard
DATASET
Create your own: use the AdventureWorks sample dataset (free from Microsoft) or Superstore dataset from Tableau's sample data
BUSINESS QUESTION
"How is revenue trending versus budget? Which product lines and regions are over/under-performing? What is the YoY change and what is driving it?"
DELIVERABLES
- Power BI report (.pbix file published to Power BI Service, or screenshots in README)
- DAX measures: Total Revenue, Revenue vs Budget %, YoY %, MTD Revenue, Top 5 Products (RANKX)
- Report pages: Executive Summary (KPI cards + trend), Regional Breakdown (map + bar), Product Performance (matrix with conditional formatting)
- At least one SWITCH-based dynamic metric selector
WHAT IT PROVES TO A HIRING MANAGER · Power BI is the most-requested BI tool in Delhi NCR job descriptions. A working .pbix with real DAX measures is the strongest possible evidence of Power BI competency.
README TIP · Share the .pbix file in the repo. Include screenshots of each report page. List all DAX measures with a one-line explanation of each.
Project 05Intermediate–Advanced⏱ 2 weeks · 🛠 Python (Pandas + openpyxl or XlsxWriter)
Python Automated Monthly Report
DATASET
Use the e-commerce or sales dataset from Project 1 or 3. The focus here is the automation, not the dataset.
BUSINESS QUESTION
"Can you build a Python script that takes a raw sales CSV and automatically produces a formatted monthly Excel report with pivot tables, calculated columns, and conditional formatting?"
DELIVERABLES
- Python script (not just a notebook): run it from command line, takes a CSV path as argument
- Output: multi-sheet Excel file — Monthly Summary, By Region, By Product, Clean Data
- Conditional formatting applied: red for below-target rows, green for above
- Email draft section: script prints a plain-text executive summary at the end
WHAT IT PROVES TO A HIRING MANAGER · This is the project that makes a hiring manager think "this analyst would save us 3 hours of manual work per week." Automation projects are remembered. Most other candidates submit dashboards; you submit a tool.
README TIP · Show the input (raw CSV columns) and output (screenshot of the Excel report). Include a usage section: python report.py --input sales_october.csv --output report_october.xlsx
Project 06Advanced⏱ 2 weeks · 🛠 Python (Pandas + SciPy + Seaborn)
A/B Test Analysis
DATASET
Kaggle: "A/B Testing" dataset by Ziad Saadi (advertising click data, control vs treatment group)
BUSINESS QUESTION
"Did the new ad design significantly increase click-through rate? Is the result practically significant (not just statistically significant)?"
DELIVERABLES
- Python notebook: data cleaning, group split validation, conversion rate by group
- Statistical test: two-proportion z-test or chi-square test using scipy.stats
- Confidence intervals for both groups
- Power analysis: what sample size would be needed to detect a 1% lift with 80% power?
- Written conclusion: would you recommend shipping the new design? Why?
WHAT IT PROVES TO A HIRING MANAGER · Statistical thinking and A/B testing are explicit requirements in product analytics, e-commerce, and marketing analyst roles. This project puts you in a separate category from analysts who only know aggregations. It directly proves you can evaluate experiments — a core skill at ₹10+ LPA roles.
README TIP · State your null hypothesis at the top. Show p-value, confidence interval, and effect size. End with a clear ship/no-ship recommendation with reasoning.
GitHub Portfolio Structure
Each project should be its own GitHub repository. Your GitHub profile itself is also a landing page — fill in your bio, pin your best 4–6 repos, and add a profile README.
# Repository structure for each project
ecommerce-sales-analysis/
├── README.md ← business question + findings + screenshots
├── data/
│ ├── raw/ ← original downloaded dataset
│ └── cleaned/ ← processed output
├── notebooks/
│ └── analysis.ipynb ← Jupyter notebook or Python script
├── sql/
│ └── queries.sql ← all SQL queries with comments
└── dashboard/
└── screenshots/ ← Power BI or Excel screenshots
✅ Write the README in plain English, not jargon
Lead with the business question and your key finding. Then describe what you did. A hiring manager should understand your project without opening a single file.
✅ Add screenshots to every README
Include at least one chart or dashboard screenshot. Hiring managers review portfolios on their phone while commuting. A visual in the README is seen; a chart inside a notebook is not.
✅ Never upload confidential or personal data
Only use public datasets (Kaggle, government open data, sample databases). Check your .gitignore before pushing — never commit .env files, API keys, or any data with real personal information.
✅ Pin your best work, not your first work
You can create 10 repos but only show the 4–6 best on your GitHub profile page. Pin the ones that demonstrate the broadest and most relevant skills.
5 Portfolio Mistakes That Get You Rejected
❌ No business question
A portfolio project titled "Data Cleaning of Sales Data" with no stated business question tells a hiring manager nothing about your thinking. Every project must open with: what were you trying to find out, and why does it matter?
❌ Only screenshots, no working code
If your GitHub shows only JPEGs of dashboards with no SQL, Python notebooks, or .pbix files, the hiring manager cannot verify your work. Always include the actual files.
❌ Tutorial datasets with tutorial analysis
Projects using the Titanic or Iris datasets from tutorials look like coursework, not original work. Use any of the six datasets in this chapter instead — they are public, but the analysis you do on them is yours.
❌ No README or a one-line README
"This is my sales project" is not a README. A useful README includes: business question, dataset source, tools used, 3–5 key findings, and a screenshot. Write it as if explaining the project to a colleague who does not know anything about it.
❌ Too many projects, all shallow
Six projects each with one chart and one SQL query is weaker than three projects with complete analyses, proper READMEs, and clear business recommendations. Finish each project properly before starting the next.
Build Your Portfolio at EVIKA ACADEMY
Our data analytics course in Noida Sector 51 guides you through building a portfolio as you learn — SQL, Python, Power BI projects on real datasets with mentor feedback. Online & offline batches available.
📱 Book Free Demo Class →Frequently Asked Questions
How many projects should a data analyst portfolio have?
Three to five well-executed projects are sufficient. The minimum is three: SQL, Python/Pandas, and a dashboard. Each must include working code, a business-question README, and at least one visual. More projects only help if each is complete and polished.
What datasets should I use for my data analyst portfolio?
Use public datasets: Kaggle (Olist e-commerce, IBM HR Attrition, A/B Testing), Northwind (SQL), AdventureWorks/Superstore (Power BI), or data.gov.in for India-specific data. Avoid Titanic and Iris — they signal coursework, not original analysis.
Do I need a GitHub account for my data analyst portfolio?
Yes — GitHub is the standard platform. Each project should be a separate repository with a README, dataset link, SQL files, Python notebooks, and dashboard screenshots. Pin your best 4–6 repos on your profile.
What is the most impressive portfolio project for a data analyst fresher?
The Python automated report project stands out for demonstrating practical value. The A/B testing project stands out at product companies. For most generalist roles: one solid SQL project (window functions + CTEs) + one Power BI dashboard with real DAX is the strongest pair.
Where can I build a data analyst portfolio with guidance in Noida?
EVIKA ACADEMY in Noida Sector 51 guides students through SQL, Python, and Power BI portfolio projects during the course with instructor feedback. Online and offline batches. WhatsApp +91-8081035456.