An interactive machine learning web application built using Streamlit that allows users to estimate California house prices by dynamically adjusting real-world features such as income, rooms, house age, population, and location.
Behind the scenes, the app trains and compares multiple ML models and explains their behavior in simple, beginner-friendly language.
- Adjust house features using sliders:
- Median income
- House age
- Number of rooms & bedrooms
- Population & occupancy
- Latitude & longitude
- Get instant price predictions in dollars
- View predictions from multiple models side-by-side
The app trains and evaluates:
- Linear Regression – simple baseline model
- Ridge Regression – regularized linear model
- Random Forest Regressor – non-linear ensemble model
- Average Ensemble – combined prediction for stability
Evaluation Metrics Used
- Mean Absolute Error (MAE)
- Root Mean Squared Error (RMSE)
- R² Score
Visual comparisons help users understand which model performs best and why.
- House price distribution visualization
- Correlation heatmap between features
- Feature importance from Random Forest
- Plain-English explanations of how each factor influences price
This helps users build intuition about housing economics, not just ML metrics.
- End-to-end machine learning pipeline
- Proper preprocessing using
PipelineandColumnTransformer - Model evaluation and ensemble techniques
- Translating ML outputs into a user-friendly product
- Streamlit best practices with caching and clean UI
- Python
- Streamlit
- Scikit-learn
- Pandas, NumPy
- Matplotlib, Seaborn
pip install -r requirements.txt
streamlit run app.pyUses the California Housing Dataset from scikit-learn, based on 1990 U.S. Census data. Target variable: Median house value (converted to USD).
Median income is the strongest driver of house prices
Tree-based models capture complex patterns better than linear ones
Combining models (ensemble) leads to more balanced predictions
Interactive tools make ML more intuitive and accessible
ML beginners looking to understand regression models
Product managers exploring feature impact
Anyone curious about how housing factors affect prices
