05AI / LLM·2022
AI House Price Prediction
Neural networks for property valuation, deployed as a live web app.
- +20%
- Accuracy vs baseline
- 10K+
- Listings trained on
- Live
- Web app
A property-valuation model trained on real listings and shipped as something you can actually use, not a notebook. The goal was a measurable improvement over the obvious baseline — and proving it.
Approach
- Trained neural networks on 10K+ property listings and tuned hyperparameters for accuracy.
- Deployed the model as a Streamlit web app for real-time valuations by end users.
- Benchmarked rigorously against baseline regression to validate that the added complexity earned its keep.