Inspiration
We were inspired by enterprise tools like SAP Analytics, but wanted to integrate MIT's "15-Minute City" research to show how hyper-local amenity access drives real estate value.
What it does
AssetVision is a machine-learning platform that predicts multifamily RevPAR growth using 10, 15, and 30-minute drive-time data. It combines quantitative forecasting with Generative AI to write strategic investment memos explaining the "why" behind the data.
How we built it
We built a Python pipeline using Streamlit for the frontend, Scikit-Learn for Random Forest regression, and the Google Gemini API (google-genai) to act as our automated investment strategist.
Challenges we ran into
Our biggest hurdle was standardizing the inconsistent amenity data across different drive-time radii and figuring out how to successfully pipe raw "feature importance" data into the LLM for analysis.
Accomplishments that we're proud of
We are proud of moving beyond a basic script to a polished, "Premium Minimalist" UI that automates the entire workflow—from raw data ingestion to final scoring—in a single click.
What we learned
We learned how to leverage Generative AI not just for chat, but as an analytical layer that translates complex mathematical models into human-readable financial strategy.
What's next for AssetVision
We plan to add geospatial visualizations for the drive-time zones and expand the model to predict performance for Industrial and Retail asset classes to generate further alpha.
Log in or sign up for Devpost to join the conversation.