Asterisim

Inspiration

We were inspired by the gap between having access to data and actually understanding it. Valuable patterns can remain hidden in spreadsheets, and uncovering them often requires statistical knowledge and experience with data analysis. We wanted to make exploring data more accessible so people could focus less on figuring out how to analyze their data and more on understanding what it reveals.

What it does

Asterisim transforms raw CSV data into interactive visualizations that reveal patterns and clusters. It helps users explore relationships within their data, understand how groups differ through statistical summaries and plain-English explanations, and identify meaningful findings. Our goal is to go beyond visualization by helping users determine what to investigate next.

How we built it

We built Asterisim around a machine-learning analysis pipeline that processes uploaded datasets and identifies clusters of similar records. We combined the analysis results with interactive data visualizations and natural-language explanations to make complex findings easier to understand. Our approach brings data processing, pattern discovery, visualization, and interpretation together in one application.

Challenges we ran into

One of our biggest challenges was integrating different implementations into a cohesive application while preserving the strengths of each. We also had to think beyond simply displaying clusters: visualizations need to communicate meaningful differences, and generated explanations must accurately reflect the underlying statistics. Balancing technical complexity, analytical accuracy, and accessibility was central to our development process.

Accomplishments that we're proud of

We're proud of building an application that aims to make machine-learning-powered data exploration more approachable. Instead of requiring users to interpret raw tables or understand clustering algorithms, Asterisim brings pattern discovery and interpretation into a single experience. We're especially excited about extending the experience from showing users what their data looks like to helping them understand what it might mean.

What we learned

We learned that useful data analysis requires more than a working machine-learning model. The results must also be understandable, the visualizations must communicate patterns clearly, and every explanation must be grounded in evidence. We also learned the importance of integrating different technical components without sacrificing usability or analytical reliability.

What's next for Asterisim

We want to develop Asterisim into a more complete data-analysis assistant. Next, we aim to improve its statistical insights and interactive visualizations, provide evidence-backed recommendations for what users should investigate next, and help users connect discovered patterns to practical actions. Ultimately, we want to make meaningful data exploration accessible to people regardless of their technical background.

Built With

• ml
• python
• react