The best data analyst portfolio project for an Indian fresher is one built on a real Indian dataset, answering a question a business or government team would genuinely ask, and presented in a way a recruiter can understand in two minutes. Generic datasets such as Titanic or Iris show that you completed a tutorial. A project on Indian rainfall, prices or public services shows that you can find messy data, clean it and say something useful about it.
Why this matters: among candidates registered on MyInternships.in, counted 1 Oct 2026, SQL is listed by 21,804 people and Python by 21,874. Listing the tool is common. Showing what you did with it is not.
Where to find the data
The Open Government Data Platform (data.gov.in) publishes thousands of datasets from central ministries and states, usually as CSV or through an API. Search by sector, check the "last updated" date and download the file with its metadata page so you can cite the source. Other public sources include state open data portals and the annual reports that ministries publish on their websites.
Before you start, read the dataset description carefully. Note the unit of each column, the time period covered and any gaps. Many public files have state names spelled differently across years, merged districts and blank rows for missing months. Cleaning those is part of the project, not a problem to hide.
Project 1: Crop prices across mandis
Question: How much do wholesale prices for one crop, such as onion or tomato, vary between markets in the same state, and in which months is the gap widest?
Steps: load daily or monthly price records, standardise market and commodity names, calculate monthly median prices per market, and compare the spread. In SQL, a window function to rank markets by price each month is a good skill to show.
Output: a line chart of the median price over time and a table of the five markets with the most volatile prices. End with two sentences on what a trader or a farmer producer organisation could do with this.
Project 2: Rainfall and reservoir levels
Question: In years with weak monsoon rainfall in a region, how quickly do reservoir levels fall afterwards?
Steps: join rainfall by subdivision with reservoir storage data by month. You will need to map reservoirs to regions, which teaches you to document assumptions. Calculate year-on-year differences and plot them.
Output: a dashboard with a map or bar chart by region and a short note listing the limitations, such as reservoirs that serve more than one region.
Project 3: Public service performance by district
Question: Which districts in a state are improving fastest on a service indicator, such as school enrolment or immunisation coverage, and which are falling behind?
Steps: clean the district list (districts are split or renamed over time, so create a mapping table), compute change over a fixed period and rank districts. Use Python with pandas for the reshaping, then load the cleaned table into Power BI or a similar tool.
Output: a ranked bar chart and a small multiple chart showing the trend for the top and bottom five. Recruiters notice when a candidate handles changing district boundaries carefully.
Project 4: Vehicle registrations and fuel type
Question: How has the share of electric or CNG vehicles in new registrations changed across states?
Steps: aggregate registrations by state, month and fuel type, then calculate shares. A clean SQL query with GROUP BY and a CASE expression for fuel categories is exactly the kind of thing interviewers ask about.
Output: a stacked area chart for India and a table for the five states with the fastest change. Add a sentence on what an automobile dealer network or a charging company might do with this view.
Project 5: Small business sales (your own collected data)
Not every project has to use government data. Ask a local shop, café or tuition centre for anonymised monthly sales or enrolment records, with permission. Remove names and phone numbers before you touch the file.
Question: Which products or batches drive most revenue, and is there a seasonal pattern?
Output: a simple Pareto chart and three recommendations. This project shows business sense, data handling and responsible use of data, which matters more now that India has a dedicated data protection law, the Digital Personal Data Protection Act, 2023.
How to present each project
Recruiters spend very little time on each portfolio link, so structure matters:
- One-line question at the top of the README or page.
- Source link and date downloaded.
- Three bullet points of findings, written for a manager, not a data scientist.
- One chart or dashboard screenshot.
- Code and SQL in folders below, with comments.
Keep notebooks tidy: remove failed experiments or move them to a separate folder.
Choosing the right tools
You do not need every tool. A strong combination for internships is SQL for querying, Excel for quick checks, Python (pandas) for cleaning and one dashboard tool. If you are new to Python, the free Python tutorial covers pandas basics. Practise query questions on the SQL interview questions page, because interviews for these roles nearly always include a live SQL question.
Using the projects when you apply
Put the two strongest projects at the top of your resume under "Projects", each with a one-line result and a link. In interviews, be ready to explain one decision you made while cleaning the data and one thing you would do differently.
When you are ready, browse data analyst internships and mention the project most relevant to each company in your application. A logistics firm will care more about the crop price project; a public policy consultancy will care more about district indicators.
A final check
Before sharing, open every link in an incognito window, check that charts have titles and units, and make sure you cite the source. A small, honest, well-documented project is more convincing than a large one that cannot be reproduced.


