A machine learning-powered tool for predicting California housing prices, built on real census data and modern ensemble methods.
97.8% R² score using Gradient Boosting — our champion model captures complex non-linear price patterns.
Trained exclusively on California housing data, capturing regional nuances from Bay Area tech hubs to Central Valley.
Compare predictions across Gradient Boosting, Random Forest, and Ridge Regression.
Tree-based models provide 95% confidence intervals so you understand the range of likely values.
Explore feature importance, regional patterns, and market drivers to understand pricing.
Deployed on Vercel with Next.js for fast, reliable predictions at scale.
Pick a city
Select from 20 California cities — area data is auto-filled.
Adjust your property
Use the sliders to set rooms, bedrooms, and household size.
See your estimate
Get both 1990 census and inflation-adjusted 2024 values instantly.
Explore insights
Visit Insights for model performance and feature importance.
Compare cities
Use the Explorer to compare prices across California.
This model is trained on the California Housing dataset from scikit-learn, derived from the 1990 U.S. Census. It contains 20,640 block-group level observations with 8 features including median income, housing age, average rooms, population, and geographic coordinates. The target variable is the median house value for each block group. While based on historical data, the model captures fundamental relationships between economic, demographic, and geographic factors that continue to drive California real estate pricing.