May 21, 2025
Upload data, run regression, get recommendations - powered by LLMs and built with analysts in mind.
LLMs are doing a fantastic job in automating repetitive and mundane day-to-day tasks. However, where they can truly add value is in performing “intelligent” tasks.
With this in mind, I started exploring how I could create “intelligent” ML models. When I say intelligent, I mean models that can provide crisp, actionable recommendations to analysts and stakeholders - not just stir through piles of data, try 10 different models, and then say “here’s everything, select whatever you like.”
To address this need, I developed an intelligent Linear Regression assistant. It’s currently hosted on Streamlit Cloud and accessible via this link: LinearLeap
This web application leverages multimodal LLMs (currently configured to use Gemini). While I’m using a free API for demonstration purposes, users can enter their own API keys to use the tool as extensively as they wish.
The Multiple Linear Regression model is still a work in progress, which I plan to enhance later. Nevertheless, I’m moderately satisfied with the Linear Regression tool’s current capabilities.
If you’re an analyst, data scientist, or stakeholder with some ML knowledge, I invite you to try it out and share your feedback. Your input will be valuable as I continue to improve the application.
GitHub repository:
LinearLeap - Github
Demo video:
(Please excuse my presentation - recording yourself is a humbling experience!!)