Housing Explorer

GRADIENT BOOSTING
Regression workspace

01 · District features→02 · Trained model→03 · Historical estimate

Explore a regression model with a visible evaluation trail. Adjust district features to see how its estimate changes.

Housing prediction, with an evaluation trail

California districts from the 1990 census. A new reproducible run compares a mean baseline, linear regression, and gradient boosting using five-fold cross-validation on training data; the selected model is evaluated once on a held-out test set.

Enter district features or try the training-set medians.

Model comparison

Model	CV RMSE (1990 dollars)

RMSE is shown in 1990-era dollars; lower is better. R² measures fit, not a percentage of correct predictions. This is a random district split, not a geographic holdout. Estimates describe historical district medians—not current home prices. Inputs far outside California districts may produce unreliable estimates. The dataset caps its highest recorded home values.

Academic project using saved model artifacts. Predictions can be imperfect; training results and limitations are shown alongside the implementation.

Compare your scenarios

Save a prediction, change the inputs, and predict again. Differences describe this model, not cause and effect.