01Why this project
Built as an independent data-product project exploring how large public datasets can be made more accessible through product design.
02The problem
Large public datasets often contain answers to important questions, but finding those answers can require significant technical knowledge.
The underlying data was useful, but difficult to explore directly. A user wanting to compare outcomes across groups would have to work through large tables and manually construct the comparisons themselves.
The goal was to make the path from question to comparison to insight much shorter.
03My role
I designed and built the dashboard end-to-end, from preparing the underlying data to deciding how the information should be presented. The project focuses on socioeconomic outcomes across SC, ST and OBC groups, using PLFS microdata as the underlying dataset.
My job wasn’t simply to visualize the data. I had to decide:
- which comparisons were useful,
- which dimensions users should be able to filter,
- how much information to show at once, and
- how to make patterns understandable without requiring users to work with the raw dataset.
04The data
1.2 million rows were not the product. The product was helping someone make sense of them.
Working with the raw dataset meant preparing and transforming the data into a structure that could support consistent comparisons across demographic, educational, employment, and geographic dimensions.
~1.2Mrows of PLFS microdata
Python, Pandasdata preparation and analysis
Plotlyinteractive visualization
~1.2M rowsone consistent structure
05From raw data to a usable interface
Rather than exposing the dataset as a collection of tables, I built the dashboard around the way someone would actually explore a question.
Clean
Prepared and transformed the data into an analysis-ready structure.
Compare
Compare outcomes across demographic groups and other dimensions.
Filter
Narrow broad results by factors such as education and location.
Explore
Interactive views make differences visible without manual analysis.
06Product thinking
The interface was designed around questions, not the dataset.
A dataset can contain hundreds of variables. That doesn’t mean a product should expose all of them. I focused the dashboard around a smaller set of comparisons that could help users investigate socioeconomic outcomes without first needing to understand the structure of the underlying data.
The interface lets a user start with a broad comparison, then progressively narrow the question through filters and views.
07What the dashboard makes visible
Rather than presenting a list of statistics, I designed the dashboard so users could change the question and see how the picture changes.
How employment outcomes differ across groups
Side-by-side comparisons make differences in employment status easier to see without manually constructing cross-tabulations.
How the picture changes with education
Filtering by education level shows whether observed differences persist, narrow, or change.
When geography changes the picture
Rural and urban breakdowns test whether a national pattern holds across settings.
08What I learned
The most interesting part of the project wasn’t finding patterns in the data. It was deciding which questions the product should make easier to answer.
Working with a dataset this large reinforced a principle I now use across my product work: